Compare commits

..
214 Commits
Author SHA1 Message Date
yiling dc8c898dae Release v0.2.2 2026-07-23 20:35:22 +08:00
Yiling-J ecc382af61 Merge pull request #218 from shumaiOne/fix/upload-task-completion-status
fix(webui): invalidate upload tasks query when upload task starts and completes
2026-07-23 20:25:44 +08:00
yiling 78dec7626b fix(webui): invalidate upload tasks query when upload task starts and completes 2026-07-23 20:18:37 +08:00
Yiling-J f87033bd3d Merge pull request #217 from shumaiOne/feat/psd-imagemagick-support
feat(transcode): add PSD support via ImageMagick and enforce sRGB conversion
2026-07-23 17:02:11 +08:00
yiling d1b0f3b64f docs: update Windows system requirements in README to include ImageMagick 2026-07-23 16:48:34 +08:00
yiling 8cdbc403db test(e2e): add PSD transcode workflow E2E test 2026-07-23 16:44:56 +08:00
yiling 90adc496dd fix(transcode): update APT package name from imagemagick-7.q16 to imagemagick 2026-07-23 16:34:01 +08:00
yiling 8d1c51abca feat(transcode): add PSD support via ImageMagick and enforce sRGB conversion 2026-07-23 16:19:25 +08:00
Yiling-J 21de21902e Merge pull request #215 from shumaiOne/feat/ai-token-usage-tracking
feat(ai-usage): add team and member token usage tracking and APIs
2026-07-23 15:26:32 +08:00
Yiling-J 330a0b99a9 Merge pull request #216 from shumaiOne/feat/webui-ai-token-usage
feat(webui): integrate AI usage API into dashboard page
2026-07-23 15:10:58 +08:00
yiling df013aa4bb feat(webui): add AI usage dashboard for team and member token tracking 2026-07-23 15:03:09 +08:00
yiling 2c04c97a02 refactor(ai-usage): use Prisma.PrismaClientKnownRequestError for P2002 handling 2026-07-23 11:35:18 +08:00
yiling f6c0a41151 fix(ai-usage): resolve code review feedback for concurrency upsert and timeframe window 2026-07-23 11:29:11 +08:00
yiling f93c8ea524 refactor(agent): invoke harness.subscribe directly without typeof check 2026-07-23 11:11:48 +08:00
yiling 783daa6d3f fix(ai-usage): align since date with top-of-hour bucket boundaries 2026-07-23 11:08:35 +08:00
yiling 1c44e5aff7 feat(ai-usage): add team and member token usage tracking and APIs 2026-07-23 11:02:54 +08:00
Yiling-J 2552d68b98 Merge pull request #214 from shumaiOne/fix/search-result-row-overflow
fix(webui): prevent horizontal overflow in search dialog result rows
2026-07-22 21:18:06 +08:00
yiling 1f714af366 fix(webui): prevent horizontal overflow in search dialog result rows 2026-07-22 21:12:45 +08:00
Yiling-J f0ca7c180b Merge pull request #213 from shumaiOne/fix/sql-query-builder-bugs
fix(search): fix buildSqlCondition date range, numeric metadata, and string parsing bugs
2026-07-22 20:43:26 +08:00
yiling 8aadaa70bf fix(search): fix buildSqlCondition date range, numeric metadata, and string parsing bugs 2026-07-22 20:33:33 +08:00
yiling 6ab650bc80 Release v0.2.1 2026-07-22 20:01:30 +08:00
Yiling-J 1886cbe420 Merge pull request #212 from shumaiOne/fix/provider-update-preserve-model-ids
fix(provider): preserve model ids on provider update
2026-07-22 16:35:08 +08:00
yiling 43dd53a30d fix(provider): preserve model ids on provider update 2026-07-22 16:27:58 +08:00
Yiling-J b1e77db6d0 Merge pull request #211 from shumaiOne/fix/pi-model-fallback
fix(agent): handle non-builtin models in getModelFromDb and upgrade pi-ai
2026-07-22 16:16:26 +08:00
yiling e9959bcc68 fix(agent): handle non-builtin models in getModelFromDb and upgrade pi-ai 2026-07-22 16:09:36 +08:00
Yiling-J fbf963fe43 Merge pull request #210 from shumaiOne/fix/generate-providers-split
refactor(core): split generated providers into per-provider files
2026-07-22 15:09:07 +08:00
yiling c364e0e695 fix(core): guard top-level sync execution and add remote fallback for model data loading 2026-07-22 12:19:37 +08:00
yiling a4b6b0b775 refactor(core): split generated providers into per-provider files 2026-07-22 12:04:07 +08:00
Yiling-J acc255fff4 Merge pull request #209 from shumaiOne/feat/skill-permission
feat(skill): add team owner skill permission control
2026-07-22 11:13:31 +08:00
yiling bc63a95bf5 fix(agent): update agent tools to throw errors directly on failure 2026-07-22 11:05:19 +08:00
yiling bbe1e7128d fix(agent): throw error on read_skill failure instead of returning error content 2026-07-22 10:43:14 +08:00
yiling 9158598c5b feat(skill): add team owner skill permission control 2026-07-22 10:27:32 +08:00
yiling b6afcc135e Release v0.2.0 2026-07-21 21:30:00 +08:00
Yiling-J fb28b7d020 Merge pull request #208 from shumaiOne/feat/agent-media-tools-refactor
feat(agent): refactor media tools to static injection with mandatory assetId and authz
2026-07-21 16:33:12 +08:00
yiling 28c3df3429 fix(agent): require mandatory userId for media tools and optimize authz 2026-07-21 16:22:14 +08:00
yiling 93e957fa4f feat(agent): refactor media tools to static injection with mandatory assetId and authz 2026-07-21 16:09:39 +08:00
Yiling-J 1ac0d1452e Merge pull request #207 from shumaiOne/fix/version-stack-agent-chat
fix(agent): pass latest version asset from version stack to prompt builder
2026-07-21 15:16:55 +08:00
yiling 5855134236 fix(agent): pass latest version asset from version stack to prompt builder 2026-07-21 15:04:38 +08:00
Yiling-J 581a12bdc9 Merge pull request #206 from shumaiOne/fix/latest-children-trashed-files
fix(asset): exclude soft-deleted children from latestChildren preview
2026-07-21 11:33:37 +08:00
yiling 851e5d975c fix(asset): exclude soft-deleted children from latestChildren preview 2026-07-21 11:19:20 +08:00
Yiling-J 6e69fc451b Merge pull request #205 from shumaiOne/fix/include-skill-id-in-prompt
fix(agent): include skill id in formatSkillsForPrompt
2026-07-21 10:46:39 +08:00
yiling b91033495c fix(agent): include skill id in formatSkillsForPrompt 2026-07-21 10:42:01 +08:00
Yiling-J a2895cd5d1 Merge pull request #204 from shumaiOne/feat/chat-context-asset-change
feat(chat): track context asset changes and update agent context
2026-07-21 10:13:09 +08:00
yiling e3352232fe feat(chat): track context asset changes and update agent context 2026-07-21 10:01:22 +08:00
Yiling-J 6d34b13266 Merge pull request #203 from shumaiOne/feat/markdown-proxy
feat(transcode): add markdown file to pdf proxy support
2026-07-20 18:37:54 +08:00
yiling 208eb6898a feat(transcode): add markdown file to pdf proxy support 2026-07-20 18:35:16 +08:00
Yiling-J b1d9c1b31b Merge pull request #202 from shumaiOne/add-fonts-noto-cjk-to-dockerfile
feat(docker): install fonts-noto-cjk in runner stage
2026-07-20 16:36:18 +08:00
yiling c016d1dffe feat(docker): install fonts-noto-cjk in runner stage 2026-07-20 16:35:19 +08:00
Yiling-J 7cd8a5bef7 Merge pull request #201 from shumaiOne/fix/transcode-pdf-font
fix(transcode): mark pdfkit as external and extract for docker runtime
2026-07-20 16:16:05 +08:00
yiling 3274d19574 fix(transcode): mark pdfkit as external and extract for docker runtime 2026-07-20 16:12:59 +08:00
Yiling-J 047174c13d Add Discord link to README 2026-07-20 15:05:43 +08:00
Yiling-J d9d7b52758 Merge pull request #200 from shumaiOne/feat/use-proxy-type
feat(asset): replace mediaType/mimeType logic with proxyType
2026-07-20 14:59:39 +08:00
yiling 52edcd1a79 fix(transcode): fix unused import and set proxyType in transcode workflow tests 2026-07-20 14:50:43 +08:00
yiling 3afe0fc3c3 feat(api): remove mediaType and mimeType from DTO responses and DB MediaInfo interface 2026-07-20 14:37:17 +08:00
yiling 5d4020dc2d fix(mime): restrict pdf proxyType strictly to pdf, csv, and txt files 2026-07-20 11:50:19 +08:00
yiling 4ad0df5876 fix(agent): block non-pdf assets in read_pdf_pages and remove mediaType fallback in AgentChatPromptBuilder 2026-07-20 11:01:22 +08:00
yiling fea153e94f fix(asset): resolve DTO schema and test typecheck errors for proxyType 2026-07-20 10:46:20 +08:00
yiling f158d04686 feat(asset): replace mediaType/mimeType logic with proxyType 2026-07-20 10:22:36 +08:00
Yiling-J c45a1d9ee9 Merge pull request #199 from shumaiOne/feat/txt-csv-pdf-proxy
feat(transcode): add PDF proxy support for TXT and CSV files
2026-07-20 09:41:28 +08:00
yiling 29995a64fd fix(transcode): add fontName to Linux TTC fonts, Windows CJK paths, and Droid fallback 2026-07-20 09:35:49 +08:00
yiling 1970535aa6 fix(transcode): use csv-parse for CSV parsing and pin pdfkit exact version 2026-07-19 22:46:13 +08:00
yiling 7574dd56e4 feat(db): add backfill migration for media.proxyType 2026-07-19 21:21:31 +08:00
yiling 081dbf6d39 feat(transcode): add PDF proxy support for TXT and CSV files 2026-07-19 21:19:43 +08:00
Yiling-J 817889e501 Merge pull request #197 from shumaiOne/feat/split-transcode-workflows
feat(transcode): split monolithic transcodeMedia into focused workflows
2026-07-19 15:01:38 +08:00
yiling aa572bf2d6 docs(transcode): add deprecation notice to legacy transcodeMedia workflow 2026-07-19 14:50:31 +08:00
yiling d9ed30fedb ci: install poppler-utils and ffmpeg system dependencies for workflow E2E tests 2026-07-19 14:43:04 +08:00
yiling 7d7a1b61c0 test(e2e): add E2E test for renderPdfPagesWorkflow 2026-07-19 13:27:07 +08:00
yiling cbe622dfeb feat(transcode): split monolithic transcodeMedia into focused workflows 2026-07-19 12:32:08 +08:00
Yiling-J b72a9435e7 Merge pull request #196 from shumaiOne/feat/agent-read-pdf-pages
feat(agent): add read_pdf_pages tool and PDF page context support
2026-07-19 12:14:20 +08:00
yiling 759e37915b refactor(transcode): add overlayAnnotationsOnBuffer and add pixel verification tests 2026-07-19 12:03:23 +08:00
yiling e2523e1a81 perf(transcode): optimize PDF page rendering pipeline using in-memory sharp buffers 2026-07-19 10:18:50 +08:00
yiling 6064a1449a fix(agent): enforce page range validation and refine pdftoppm error check 2026-07-19 10:17:21 +08:00
yiling f68ba70656 feat(agent): add read_pdf_pages tool and PDF page context support 2026-07-19 09:45:14 +08:00
Yiling-J 68a0fb7319 Merge pull request #195 from shumaiOne/feat/pdf-worker-route
feat(web): register pdf.worker route in production
2026-07-18 19:44:43 +08:00
yiling 661ec06cef fix(webui): ensure pdf.worker path is root-relative 2026-07-18 19:14:56 +08:00
yiling 6000b8e366 feat(web): register pdf.worker route manually in prod mode 2026-07-18 19:10:07 +08:00
Yiling-J db3551de1d Merge pull request #193 from shumaiOne/feat/pdf-support
feat(transcode): add PDF preview, sprite generation, and annotation support
2026-07-18 15:42:15 +08:00
yiling dc20c7f548 test(e2e): add PDF workflow E2E test case using test.pdf fixture 2026-07-18 15:33:05 +08:00
yiling f76e7a8e8d fix(webui): cancel renderTask immediately if active flag is false before or during render 2026-07-18 15:17:38 +08:00
yiling cfdc085ae7 chore(deps): move pdfjs-dist to webui package.json and clean root dependencies 2026-07-18 15:15:45 +08:00
yiling 779706a3e7 test(core): add unit tests for generatePdfSprite and getPdfInfo in TranscodeService 2026-07-18 15:07:28 +08:00
yiling 06dc929a2a fix(webui): destroy loadingTask and cancel renderTask on unmount 2026-07-18 14:05:10 +08:00
yiling 0ad3aca083 fix(webui): add type declaration for *.mjs worker imports 2026-07-18 14:03:31 +08:00
yiling b59f439683 feat(webui): self-host pdf.worker.min.mjs in public directory 2026-07-18 13:08:55 +08:00
yiling d270f3f62c fix(webui): add keyboard navigation shortcuts for PDF viewer 2026-07-18 11:25:59 +08:00
yiling ea201ad19f feat(webui): support custom timestamp formatting for PDF viewer comments (P9) 2026-07-18 11:20:13 +08:00
yiling 2b72f3038a fix(webui): enable hasTimestamp for PDF viewer comments configuration 2026-07-18 10:55:29 +08:00
yiling c1ddbe4be7 fix(webui): sync pdf page with seekTo and onTimeUpdate 2026-07-18 10:49:53 +08:00
yiling 3ada194a8e fix(asset): presign poster thumbnail for PDF and poster-enabled assets 2026-07-18 10:41:28 +08:00
yiling bc1491bed5 feat(transcode): add PDF preview, sprite generation, and annotation support 2026-07-18 10:32:38 +08:00
yiling 3383079787 Release v0.1.10 2026-07-16 16:48:25 +08:00
Yiling-J f888a575b2 Merge pull request #192 from shumaiOne/feat/disable-agent-tools
feat(agent): support disabling specific tools for agents
2026-07-16 16:38:32 +08:00
yiling 0a7d5d1f46 feat(agent): support disabling specific tools for agents 2026-07-16 15:56:56 +08:00
Yiling-J 879922a9fe Merge pull request #191 from shumaiOne/feat/comment-delete-video-shortcuts
feat(comments): support comment deletion and video player shortcuts
2026-07-16 11:16:18 +08:00
yiling 58967d1a63 minor ui improvement 2026-07-16 11:06:26 +08:00
yiling e978878a90 refactor(comments): unconditionally delete S3 attachment objects directly 2026-07-16 10:59:14 +08:00
yiling ecfbfc2bac fix(shortcuts): resolve keyboard accessibility and contenteditable issues 2026-07-16 10:23:47 +08:00
yiling 490c6a3da6 refactor(comments): simplify teamId lookup and clean up comment attachments in S3 and database 2026-07-16 10:17:32 +08:00
yiling 62b1fc1d37 feat(comments): support comment deletion and video player shortcuts 2026-07-16 09:55:05 +08:00
Yiling-J fa6bf73452 Merge pull request #189 from shumaiOne/feat/workflow-e2e-tests
test(e2e): add E2E workflow tests for remaining agent and transcode services
2026-07-15 15:36:41 +08:00
yiling 25c747aa87 test(e2e): add E2E workflow tests for remaining agent and transcode services 2026-07-15 15:19:25 +08:00
Yiling-J ff7712744d Merge pull request #188 from shumaiOne/feat/workflow-e2e-tests
feat(workflow-e2e): add E2E tests for agentAutofillMedia workflow
2026-07-15 14:45:04 +08:00
yiling 6f369625d8 docs(agents): update AGENTS.md with E2E workflow testing requirements 2026-07-15 12:04:45 +08:00
yiling c6acd075e7 refactor(workflow-core): statically import Worker type from @temporalio/worker to ensure activeWorkers type-safety 2026-07-15 11:58:17 +08:00
yiling 7d9f3276f1 test(e2e): verify multiple types in autofill and check that non-autofill field is ignored 2026-07-15 11:54:29 +08:00
yiling d524a81ecd chore(vitest): exclude packages/e2e/** from root unit test suite 2026-07-15 11:40:57 +08:00
yiling 83c427e646 refactor(e2e): mock LLM prompt via AgentHarness prototype spy instead of global activity override 2026-07-15 10:52:01 +08:00
yiling b018fd0fec feat(workflow-e2e): add E2E tests for agentAutofillMedia workflow
- Setup E2E package under packages/e2e
- Use testcontainers to spin up postgres and temporal dev server
- Overwrite activities in global registry to mock AI model output
- Track temporal workers and expose shutdownWorkers() method for clean shutdown
- Integrate test:e2e:workflow into root package.json and GitHub actions CI
2026-07-15 10:14:59 +08:00
yiling 8efc9c1865 Release v0.1.9 2026-07-14 20:21:21 +08:00
Yiling-J 0eeb2f18f3 Merge pull request #187 from shumaiOne/fix/chat-api-db-queries
refactor(api/chat): remove direct Prisma query from API layer
2026-07-14 16:47:44 +08:00
yiling 6643d63e81 refactor(api/chat): remove direct Prisma query from API layer
API layer routes in packages/api/src/chat.ts accessed the database directly via prisma, violating communication layer boundaries.

This refactoring migrates these queries to packages/core/src/chat/chat.ts:
- Added getNewSessionMessages to fetch mapped SSE entries.
- Added getChatWorkflowStatus to check task completions.
- Added abortSession to handle workflow cancellation and task updates in core.
- Updated both core and API tests to match refactored signatures and mock boundaries.
2026-07-14 16:40:55 +08:00
Yiling-J b53b76bc6e Merge pull request #186 from shumaiOne/feat/agent-abort
feat(agent): support aborting agent execution
2026-07-14 16:35:19 +08:00
yiling 636e92af4c fix(agent): unregister temporal cancellation listener on completion 2026-07-14 16:08:43 +08:00
yiling a7dc7bbff2 perf(agent): optimize active task lookup using indexed sessionId column 2026-07-14 15:44:10 +08:00
yiling cd48d2d389 fix(agent): resolve lint, typecheck, and stop button UI issues 2026-07-14 10:35:25 +08:00
yiling 8a942d4f4e feat(agent): support aborting agent execution in local and temporal modes 2026-07-14 10:25:06 +08:00
Yiling-J e772581542 Merge pull request #185 from shumaiOne/feat/chatbot-ui-layout
feat(chat): chatbot UI
2026-07-13 16:55:25 +08:00
yiling 7baf95e48c fix(webui): preserve chatbot sidebar scroll position across navigations 2026-07-13 16:49:04 +08:00
yiling 15cbf12162 fix(webui): render theme-aware markdown code blocks in dark/light mode 2026-07-13 14:52:38 +08:00
yiling d560032013 feat(chat): send SSE keep-alive ping heartbeats during polling 2026-07-13 12:17:00 +08:00
yiling 02f2fde8ac fix(chat): remove explicit any in test mock to fix CI lint 2026-07-13 11:52:44 +08:00
yiling e080167ac5 feat(chat): enforce team-scoped authorization and session creator check
- Update chat API routes and service methods to include and verify teamId context
- Validate that the requesting user is the creator of the chat session during continuation
- Update frontend chatbot store and components to pass teamId parameter
- Update backend API/core and frontend store tests to verify team scoping and creator checks
2026-07-13 11:35:55 +08:00
yiling 61c49b4d7f feat(chat): allow user to choose chatbot in sidebar
- Remove chatbotAgentId from team settings and schema definition
- Make agentId a required parameter in ChatRequest DTO and chatService
- Fetch chatbot agents on chatbot sidebar mount and default to first enabled chat agent
- Render an agent selector dropdown inside the message input area
- Update existing chat sessions to follow the selected agent when continuing a conversation
2026-07-13 10:10:56 +08:00
yiling bad93466ae feat(webui): use filled bot icon when chatbot sidebar is open 2026-07-13 09:37:37 +08:00
yiling 1f8b0bac21 update ui 2026-07-13 09:34:51 +08:00
yiling c55380c0fd feat(webui): make toolCall block collapsible and render footer conditionally 2026-07-12 15:40:21 +08:00
yiling 3de57fd176 feat(webui): refactor chatbot message rendering to type-safe nested switch 2026-07-12 15:18:31 +08:00
yiling 0a4622ed8a syle minor fix 2026-07-12 10:29:29 +08:00
yiling bcb17acaf7 chore(serve): set idleTimeout to 120 seconds in Bun.serve 2026-07-11 21:11:18 +08:00
yiling b57a3e0239 chore(chat): support isNewChat payload flag and fix any casts 2026-07-11 21:05:47 +08:00
yiling c5ce4d07e9 feat(chat): add chatbot session name generation 2026-07-11 20:54:09 +08:00
yiling b5d52a4e56 feat(chat): add AgentSessionType enum and filter chatbot history sessions 2026-07-11 20:14:40 +08:00
yiling 94a3ec0b9b feat(chat): support optimistic context display entry locally during message sending 2026-07-11 17:02:58 +08:00
yiling bd897b951a chore: format codebase 2026-07-11 16:58:37 +08:00
yiling 482df3315c feat(chat): append and display context_display_info custom entry for attached assets 2026-07-11 16:58:09 +08:00
yiling cc9b32716d feat(chat): filter out context messages from API list and SSE stream 2026-07-11 15:47:48 +08:00
yiling 35db3fc49e fix(chatbot): preserve historical message order during SSE playback 2026-07-11 15:12:59 +08:00
yiling 6a5f8db0d6 fix(chatbot): preserve optimistic message during history playback of active sessions 2026-07-11 15:08:07 +08:00
yiling b13c47f207 feat(chatbot): highlight entire sidebar area with semi-transparent cover on dragover 2026-07-11 14:56:21 +08:00
yiling 2f68043e17 fix(chatbot): clear attached assets on message send and remove drag-drop toast 2026-07-11 14:41:23 +08:00
yiling 3d266a5606 feat(chatbot): remove agentId from request and resolve it via team settings, toggle chatbot icon with tooltip guard 2026-07-11 11:49:38 +08:00
yiling 87d811dad6 style(chatbot): move send button inside input border area to match comment input design 2026-07-11 10:59:56 +08:00
yiling 4a7da10ca8 fix(chatbot): prevent Enter from sending message during IME composition 2026-07-11 10:55:01 +08:00
yiling d19200e9ef feat(chatbot): implement chatbot sidebar, message view, history mode and sse stream 2026-07-11 10:49:40 +08:00
yiling 77718a3e1a refactor(webui): render chatbot button as standalone Bot icon outside the toggle group 2026-07-11 10:19:41 +08:00
yiling 8d2d915f36 feat(webui): add chatbot layout and drag-drop context store 2026-07-11 10:16:46 +08:00
Yiling-J 6ede64495f Merge pull request #184 from shumaiOne/feat/app-e2e-tests
feat(e2e): set up fullstack app-level e2e test framework and signup test case
2026-07-10 23:06:16 +08:00
yiling 13f220ef0f refactor(e2e): replace S3 bucket name check with explicit SHUMAI_E2E env flag 2026-07-10 20:11:54 +08:00
yiling 4e9fae96b1 fix(ci): split E2E jobs and support skipping app server via E2E_SKIP_APP_SERVER 2026-07-10 19:41:31 +08:00
yiling 8d23d8b1c3 fix(ci): split E2E tests into harness and app jobs 2026-07-10 19:28:14 +08:00
yiling 1e1ddd1a3d fix(e2e): declare types dependencies and adjust CI projects execution using --no-deps 2026-07-10 19:10:30 +08:00
yiling 55457fcd36 fix(webui): snap playhead to last frame upon ended event in useFramePlayer 2026-07-10 16:35:41 +08:00
yiling 9414ebaefd feat(e2e): run harness tests in parallel and app E2E tests sequentially using project dependencies 2026-07-10 16:27:51 +08:00
yiling 515c491d5d fix(e2e): restore Playwright empty object destructuring for fixture parameters 2026-07-10 16:21:09 +08:00
yiling 0aba2ee557 refactor(e2e): migrate database reset logic to standard Playwright custom fixture 2026-07-10 16:18:45 +08:00
yiling cd96015b9d feat(e2e): add browser selection support for app-level E2E tests 2026-07-10 15:35:59 +08:00
yiling 26982ee494 feat(e2e): set up fullstack app-level e2e test framework and signup test case 2026-07-10 15:21:01 +08:00
Yiling-J af5c85a87b Add installation command for ffmpeg using Chocolatey 2026-07-10 09:57:21 +08:00
yiling 5636a1b581 Release v0.1.8 2026-07-10 09:39:40 +08:00
Yiling-J 27af3a315e Merge pull request #183 from shumaiOne/fix/splitting-build-error
fix(webui): enable build splitting and code-split all page routes
2026-07-09 20:14:23 +08:00
yiling d3bbd42208 fix(web): optimize serveMainHtml performance by processing HTML at startup 2026-07-09 15:50:09 +08:00
yiling c3d88bb972 docs: add concise route code-splitting guideline to AGENTS.md 2026-07-09 15:32:15 +08:00
yiling a80e9de6cc feat(webui): code-split remaining team and share sub-routes to optimize vendor chunk size 2026-07-09 15:16:33 +08:00
yiling 6b5c2d8b23 feat(webui): enable route-level manual code-splitting to optimize bundle size 2026-07-09 15:09:39 +08:00
yiling 8d19b578e7 fix(build): resolve code-splitting chunk import issue on backend and frontend 2026-07-09 15:00:23 +08:00
Yiling-J e20e147445 Merge pull request #182 from shumaiOne/update-readme-notes
docs(readme): clarify displayed file size represents original file size
2026-07-09 10:40:58 +08:00
yiling 2116f2b1ab docs(readme): clarify displayed file size represents original file size 2026-07-09 10:40:24 +08:00
Yiling-J 7889b08008 Merge pull request #180 from shumaiOne/feat/chatbot-backend-apis
feat(chatbot): implement type-safe Chat APIs with compaction support
2026-07-08 10:07:00 +08:00
yiling 10279d74bc feat(chatbot): append turn-specific agentsInstruction as a custom message session entry 2026-07-07 21:51:10 +08:00
yiling 985bf0170c fix(api): prevent SSE connection resource leak by checking stream.aborted 2026-07-07 20:59:25 +08:00
yiling f04729b4bf refactor(chatbot): use SessionTreeEntry type directly in ChatService, removing SimpleMessageEntry 2026-07-07 14:16:31 +08:00
yiling 55026af22e refactor(chatbot): import AgentMessage type statically in DTO layer 2026-07-07 12:30:41 +08:00
yiling b8eddcd854 feat(chatbot): return rich AgentMessage structure directly from listMessages API 2026-07-07 12:25:32 +08:00
yiling d326a4cab8 refactor(chatbot): make mapEntryToMessage fully type-safe without any casts 2026-07-07 12:04:08 +08:00
yiling 243a4ab2c5 feat(chatbot): remove manual AgentSessionEntry creation from ChatService 2026-07-07 11:50:31 +08:00
yiling e4c1830e02 feat(chatbot): delegate context prompt building to AgentChatPromptBuilder in workflow 2026-07-07 11:42:40 +08:00
yiling 7b6db5b56c feat(chatbot): require agentId in ChatRequest and validate agent presence 2026-07-07 11:27:29 +08:00
yiling 63590d5329 feat(chatbot): decouple chatbot sessions from AssetComment table 2026-07-07 11:12:15 +08:00
yiling ccf742e158 refactor(chatbot): make user parameter in startOrContinueChat type-safe 2026-07-07 11:04:51 +08:00
yiling e55ee3a3f0 feat(chatbot): add chatbot backend api support 2026-07-07 10:55:22 +08:00
yiling 5e73fcf4aa update readme 2026-07-07 09:32:23 +08:00
yiling ca5f54b385 update readme 2026-07-07 09:27:50 +08:00
yiling 6e460430dd Release v0.1.7 2026-07-06 13:01:47 +08:00
yiling ffc578cb45 fix(comments): prevent reload of main viewer by invalidating only comments queries 2026-07-06 12:51:25 +08:00
Yiling-J 1414632c4a Merge pull request #179 from shumaiOne/feat/comment-completion
feat(comments): allow marking comments as complete or incomplete
2026-07-06 12:50:04 +08:00
yiling a312ecf7ef style(comments): use custom filled circle check icon and make actions row always visible 2026-07-06 12:40:20 +08:00
yiling e1d3dd46db feat(comments): allow marking comments as complete or incomplete 2026-07-06 12:24:40 +08:00
yiling af7d636f31 update cn translate 2026-07-06 11:49:51 +08:00
yiling b98b710222 update doc 2026-07-06 11:22:32 +08:00
Yiling-J 79b19d3983 Revise feature descriptions in README_CN.md
Updated feature descriptions and improved clarity in the README_CN.md file.
2026-07-06 11:15:29 +08:00
yiling 474fe7ced1 fix(webui): center unread badge count 2026-07-06 11:06:09 +08:00
Yiling-J bf82eff068 Merge pull request #178 from shumaiOne/feat/i18n-fixes
feat(i18n): localize members dialog, notifications, uploads, user menus, dates, and players
2026-07-06 10:59:24 +08:00
yiling 01a9c7cde6 refactor(i18n): address code review feedback on string legos, hardcoded locales, and fragile checks 2026-07-06 10:52:55 +08:00
yiling 45f3255796 feat(i18n): localize members dialog, notifications, uploads, user menus, dates, and players 2026-07-06 10:38:41 +08:00
Yiling-J 64ad183bc1 Merge pull request #177 from shumaiOne/docs-s3-storage
docs: improve documentation and examples for S3 compatible storage
2026-07-06 10:20:31 +08:00
yiling af8bc30409 docs: improve documentation and examples for S3 compatible storage 2026-07-06 10:13:12 +08:00
Yiling-J aac8bea4e8 Merge pull request #176 from shumaiOne/fix/upload-mime-type
fix(upload): resolve empty and generic MIME types in confirmFileUpload
2026-07-05 22:10:22 +08:00
yiling a05c0ac0db fix(upload): resolve empty and generic MIME types in confirmFileUpload
Use Bun's built-in file resolver to fall back to extension-based MIME type detection when the mediaType is missing, empty, or generic ('application/octet-stream').
2026-07-05 21:55:26 +08:00
Yiling-J 714158b5b2 Merge pull request #175 from shumaiOne/feat/audio-support
feat(audio): support frame-accurate audio playback, comments, and AI safeguards
2026-07-05 19:33:37 +08:00
yiling 8ad25215dd refactor(agent): prevent creating autofill and embedding tasks for audio assets 2026-07-05 16:31:14 +08:00
yiling 73aa374b2b feat(agent): prevent exposing screenshot and analyze_image tools and executing autofill/embeddings for audio files 2026-07-05 16:26:48 +08:00
yiling cee4142936 style(webui): use dark background for audio preview to keep duration clear 2026-07-05 16:16:04 +08:00
yiling 0f6912c977 feat(webui): render duration overlay for audio files in file browser 2026-07-05 16:13:00 +08:00
yiling 868e7b2a7b fix(webui): remount file viewer when file ID changes to reset state 2026-07-05 16:10:01 +08:00
yiling ccfcdce23c fix(upload): correct .wma MIME type to audio prefix on upload and add tests 2026-07-05 15:57:29 +08:00
yiling 8d336836cc test(webui): add audio comment seek roundtrip E2E test case 2026-07-05 15:37:30 +08:00
yiling f24c8afa3d test(webui): add e2e test to verify playhead advances immediately for audio assets 2026-07-05 15:28:18 +08:00
yiling e511af219e fix(webui): immediately use currentTime fallback in useFramePlayer for audio files 2026-07-05 15:23:52 +08:00
yiling 5cae1adce1 fix(webui): hide zoom/fit controls and comment draw button for audio assets 2026-07-05 15:19:53 +08:00
yiling a2e982f2e6 feat(transcode): add audio file support with AAC stereo mp4 proxy and styled viewer 2026-07-05 15:11:19 +08:00
Yiling-J 9a7abaed1a Merge pull request #174 from shumaiOne/refactor/file-viewer-registry
refactor(webui): modularize file viewer components into registry-driven architecture
2026-07-05 12:40:41 +08:00
yiling e03c44a00e docs(webui): add file viewer registry developer documentation 2026-07-05 12:26:50 +08:00
yiling 1524d41392 fix(webui): resolve react hook order violation in video viewer 2026-07-05 12:04:35 +08:00
yiling 4cc35d3453 refactor(webui): modularize file viewer components into registry-driven architecture 2026-07-05 10:51:42 +08:00
519 changed files with 106395 additions and 22334 deletions
+70 -5
View File
@@ -32,7 +32,7 @@ jobs:
- name: Install system dependencies
run: |
sudo apt-get update
sudo apt-get install -y ripgrep bubblewrap socat
sudo apt-get install -y ripgrep bubblewrap socat ffmpeg poppler-utils imagemagick
- name: Lint
run: bun run lint
@@ -57,7 +57,7 @@ jobs:
bun run /tmp/shumai-local-release/node/node_modules/@shumai-one/shumai-transcode/bin/shumai-transcode-app.js --check
e2e_video_player:
name: E2E (${{ matrix.project }})
name: Harness E2E (${{ matrix.project }})
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
@@ -86,13 +86,78 @@ jobs:
- name: Install Playwright browser (${{ matrix.project }})
run: bunx playwright install --with-deps ${{ matrix.project }}
- name: Run video player e2e tests
run: bun run test:e2e --project=${{ matrix.project }}
- name: Run video player harness tests
env:
E2E_SKIP_APP_SERVER: true
run: bun run test:e2e --project=harness-${{ matrix.project }}
- name: Upload Playwright report
if: ${{ !cancelled() }}
uses: actions/upload-artifact@v4
with:
name: playwright-report-${{ matrix.project }}
name: playwright-report-harness-${{ matrix.project }}
path: playwright-report/
retention-days: 7
e2e_app:
name: App E2E (${{ matrix.project }})
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
include:
- os: ubuntu-latest
project: chromium
- os: ubuntu-latest
project: firefox
# App E2E needs Docker for testcontainers, run webkit on Ubuntu
- os: ubuntu-latest
project: webkit
steps:
- name: Checkout
uses: actions/checkout@v4
- name: Setup Bun
uses: oven-sh/setup-bun@v1
with:
bun-version: latest
- name: Install dependencies
run: bun install
- name: Install Playwright browser (${{ matrix.project }})
run: bunx playwright install --with-deps ${{ matrix.project }}
- name: Run app E2E tests
run: bun run test:e2e --project=app-${{ matrix.project }} --no-deps
- name: Upload Playwright report
if: ${{ !cancelled() }}
uses: actions/upload-artifact@v4
with:
name: playwright-report-app-${{ matrix.project }}
path: playwright-report/
retention-days: 7
workflow_e2e:
name: Workflow E2E
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v4
- name: Setup Bun
uses: oven-sh/setup-bun@v1
with:
bun-version: latest
- name: Install dependencies
run: bun install
- name: Install system dependencies
run: |
sudo apt-get update
sudo apt-get install -y ripgrep bubblewrap socat ffmpeg poppler-utils imagemagick
- name: Run Workflow E2E Tests
run: bun run test:e2e:workflow
+1 -1
View File
@@ -48,7 +48,7 @@ jobs:
- name: Install system dependencies
run: |
sudo apt-get update
sudo apt-get install -y ripgrep bubblewrap socat libcairo2-dev libpango1.0-dev libjpeg-dev libgif-dev librsvg2-dev fd-find
sudo apt-get install -y ripgrep bubblewrap socat ffmpeg poppler-utils imagemagick libcairo2-dev libpango1.0-dev libjpeg-dev libgif-dev librsvg2-dev fd-find
- name: Install dependencies
run: bun install
+1
View File
@@ -11,6 +11,7 @@ data/
*.db
*.diff
*.orig
.agents
# Playwright
/playwright-report/
+17 -1
View File
@@ -91,6 +91,7 @@ Add any additional context, caveats, or follow-up work.
- `bun run typecheck`
- `bun run test`
- `bun run test:e2e`
- `bun run test:e2e:workflow`
- **Backend Testing Mandate**: Every backend feature, service method, workflow, and activity MUST be accompanied by comprehensive tests. Logic-heavy code without corresponding test coverage is considered incomplete.
@@ -140,7 +141,7 @@ We follow Bun's monorepo conventions for dependency management:
3. **Local DevDependencies**: Tools specific to a single package (e.g., `@vitejs/plugin-react` for `webui`) should be declared in that package's local `package.json`.
4. **Workspace Imports**: Use `workspace:*` for internal package dependencies.
5. **Clean Root**: The root `package.json` must not contain runtime `dependencies`. It is reserved for shared `devDependencies` and workspace-wide `scripts`.
6. **Installation**: Always run `bun install` from the root. Use the `--filter` flag if you only want to update a specific package (e.g., `bun install --filter @shumai/webui`).
6. **Adding Packages**: Never edit `package.json` manually to add new dependencies. Always use `bun add` (e.g., `bun add <package> --filter <workspace>`) from the root without specifying an explicit version, letting Bun resolve the correct version.
## TypeScript Configuration
@@ -197,10 +198,15 @@ The frontend is built with:
- **Prefer System Colors**: Always prefer using semantic system color design tokens (e.g., `bg-background`, `text-foreground`, `text-sidebar-primary`, `text-muted-foreground`, etc.) over hardcoded color utilities (e.g., `bg-blue-50`, `text-emerald-600`, `bg-emerald-500`). This ensures seamless compatibility with light and dark themes.
### Route Code-Splitting
To optimize initial bundle size, all routes must be code-split into `[route].tsx` (lightweight route definition and loaders) and `[route].lazy.tsx` (UI component and heavy imports).
### Development
- **Run Dev**: `bun run dev` (Runs on port 3000)
## Testing
We use **Vitest** for testing (via `bun run test`). **Do not use `bun test`** as it uses Bun's native test runner which is not compatible with this project's test suite.
@@ -268,6 +274,14 @@ describe('Team API', () => {
})
```
### Workflow E2E Tests
- Located in `packages/e2e/workflow/**/*.test.ts`.
- Run via `bun run test:e2e:workflow` (which executes tests against both `local` and `temporal` executors dynamically).
- **Mandatory test coverage**: When creating or updating workflows or activities, you MUST create or update the corresponding E2E tests.
- **Mocking**: Mock ONLY AI API calls. Do not mock S3, databases, or media/transcode extraction services.
- Mock the AI response by spying on `AgentHarness.prototype.prompt`:
## Prisma Configuration & Migrations
- Schema: `packages/db/prisma/schema.prisma`
@@ -300,6 +314,8 @@ We use `prisma-json-types-generator` to enforce strict type-safety for Prisma `J
- Development: Use `bun --bun run prisma migrate dev` to create and apply migrations during development.
- Production: Use `bun --bun run prisma migrate deploy` to apply pending migrations in production environments.
- **No Manual Migration Creation**: Never create migration SQL files or directories manually by hand. Always use Prisma CLI commands (e.g. `bun --bun run prisma migrate dev --create-only` to generate a migration template, or `bun --bun run prisma migrate dev`) so Prisma correctly tracks migration metadata and checksums.
- **No Automatic Dev DB Reset**: Do not run `prisma migrate reset --force` or commands that force-reset the database automatically. If migrations become out of sync or a reset is required, stop executing, report the situation to the user, and present suggested manual cleanup/reset steps for the user to execute.
### Commands
+3
View File
@@ -49,6 +49,9 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
ca-certificates \
bubblewrap \
gosu \
poppler-utils \
imagemagick \
fonts-noto-cjk \
&& apt-get clean \
&& rm -rf /var/lib/apt/lists/*
+98 -18
View File
@@ -9,7 +9,7 @@
</p>
<p align="center">
<a target="_blank" href="https://shumai.one/">Website</a> | <a target="_blank" href="https://docs.shumai.one/introduction">Docs</a> | <a target="_blank" href="https://staging.shumai.one">Demo</a> | <a target="_blank" href="README_CN.md">中文</a>
<a target="_blank" href="https://shumai.one/">Website</a> | <a target="_blank" href="https://docs.shumai.one/introduction">Docs</a> | <a target="_blank" href="https://staging.shumai.one">Demo</a> | <a target="_blank" href="https://discord.gg/sUzhujK6DD">Discord</a> | <a target="_blank" href="README_CN.md">中文</a>
</p>
![Shumai App Screenshot](docs/screenshot.webp)
@@ -32,7 +32,7 @@
- **Semantic Search**: Find assets based on visual or conceptual search queries using vector embeddings(Gemini Embedding 2).
> [!NOTE]
> **Media Processing & Storage:** When uploading media, Shumai always preserves and stores your original raw file. For all media types (both images and videos), Shumai creates a 480p WebP poster image to serve as the preview in the file list view. Additionally:
> **Media Processing & Storage:** When uploading media, Shumai always preserves and stores your original raw file. Note that the file size shown in the system represents the original file size and does not include the size of any transcoded files. For all media types (both images and videos), Shumai creates a 480p WebP poster image to serve as the preview in the file list view. Additionally:
> * **For Images:** Shumai generates a high-quality WebP proxy at the exact same resolution as your original file for optimized viewing.
> * **For Videos:** It generates at least one proxy video (depending on your transcoding settings), alongside a sprite image and a small preview video (capped at 1600 frames) to enable fast timeline scrubbing and lightning-quick previews.
@@ -57,14 +57,26 @@ Docker Compose is the fastest way to get Shumai running. You do not need to clon
3. Configure environment variables (optional):
* `SHUMAI_SERVER_PORT` controls the port the Shumai server listens on. The default is `3000`.
* By default, Shumai uses the bundled local storage service in this Docker Compose setup. To use an external S3-compatible storage instead, set `AWS_ENDPOINT_URL_S3` to the provider's endpoint URL and configure the corresponding AWS credentials and bucket settings as needed.
* By default, Shumai uses the bundled local storage service in this Docker Compose setup. To use an external S3-compatible storage instead, configure the environment variables for your S3 provider. Here are examples for AWS S3 and Cloudflare R2:
**AWS S3 Example:**
```yaml
STORAGE_BACKEND: s3
S3_BUCKET: your-bucket-name
S3_ACCESS_KEY_ID: your-access-key-id
S3_SECRET_ACCESS_KEY: your-secret-access-key
AWS_ENDPOINT_URL_S3: https://s3.us-east-1.amazonaws.com
```
**Cloudflare R2 Example:**
```yaml
STORAGE_BACKEND: s3
S3_BUCKET: your-bucket-name
S3_REGION: auto
S3_ACCESS_KEY_ID: your-access-key-id
S3_SECRET_ACCESS_KEY: your-secret-access-key
AWS_ENDPOINT_URL_S3: https://<account-id>.r2.cloudflarestorage.com
```
AWS_ENDPOINT_URL_S3: S3_ENDPOINT_URL
S3_REGION: S3_REGION
S3_BUCKET: BUCKET_NAME
S3_ACCESS_KEY_ID: ID
S3_SECRET_ACCESS_KEY: KEY
```
* By default, `AWS_ENDPOINT_URL_S3` is set to: `http://localhost:{SHUMAI_SERVER_PORT}`, if you use local storage and expose Shumai on a custom host/port combination, set `AWS_ENDPOINT_URL_S3` to the external URL that browsers will use to upload files.
For example, if you change the port mapping in `docker-compose.yaml` from `3000:3000` to `12345:3000` and deploy on a server with IP address `12.34.56.78`, set:
@@ -108,23 +120,91 @@ mkdir shumai && cd shumai
#### Step 3: Install Platform-Specific Dependencies
Make sure the following system dependencies are installed on your host machine before installing Shumai:
#### Linux
| Package | Description | Ubuntu/Debian | Fedora | Arch | macOS |
| :------------- | :------------------------------------------------------ | :------------------------------- | :------------------------------- | :-------------------------------------- | :--------------------- |
| **ffmpeg** | Media transcoding and metadata extraction | `sudo apt install -y ffmpeg` | `sudo dnf install -y ffmpeg` | `sudo pacman -S --noconfirm ffmpeg` | `brew install ffmpeg` |
| **bubblewrap** | Sandboxing environment for secure AI agent execution | `sudo apt install -y bubblewrap` | `sudo dnf install -y bubblewrap` | `sudo pacman -S --noconfirm bubblewrap` | _NOT REQUIRED_ |
| **socat** | Bidirectional socket relay for sandbox network bridging | `sudo apt install -y socat` | `sudo dnf install -y socat` | `sudo pacman -S --noconfirm socat` | `brew install socat` |
| **ripgrep** | Fast search tool for workspace security policies | `sudo apt install -y ripgrep` | `sudo dnf install -y ripgrep` | `sudo pacman -S --noconfirm ripgrep` | `brew install ripgrep` |
Shumai requires the following system packages on Linux:
- **`ffmpeg`** – Used for media transcoding and metadata extraction.
- **`poppler`** (`poppler-utils`) – Used for PDF page image extraction and PDF sprite preview generation (`pdftoppm`).
- **`imagemagick`** – Used for PSD format transcoding and color profile conversion to sRGB.
- **`bubblewrap`**, **`socat`**, and **`ripgrep`** – Required by the AI agent sandbox (`anthropic-experimental/sandbox-runtime`) for process isolation, networking, and workspace search.
Install all required packages with one command:
> Ubuntu/Debian
```bash
sudo apt install -y ffmpeg poppler-utils imagemagick bubblewrap socat ripgrep
```
> Fedora
```bash
sudo dnf install -y ffmpeg poppler-utils ImageMagick bubblewrap socat ripgrep
```
> Arch Linux
```bash
sudo pacman -S --noconfirm ffmpeg poppler imagemagick bubblewrap socat ripgrep
```
> [!NOTE]
> **Ubuntu 24.04+ Note:** These releases restrict unprivileged user namespaces by default. To allow `bubblewrap` and the sandbox isolation layer to function, disable this restriction:
> **Ubuntu 24.04+** restricts unprivileged user namespaces by default. This prevents `anthropic-experimental/sandbox-runtime` from using `bubblewrap` for sandbox isolation.
>
> To temporarily allow user namespaces:
>
> ```bash
> sudo sysctl -w kernel.apparmor_restrict_unprivileged_userns=0
> ```
>
> Alternatively, configure an AppArmor profile that grants user namespace creation (`userns`) privileges to the relevant binaries.
> Alternatively, configure an AppArmor profile that grants the required `userns` permission to the relevant binaries.
---
#### macOS
Shumai requires:
- **`ffmpeg`** – Used for media transcoding and metadata extraction.
- **`poppler`** – Used for PDF page image extraction and PDF sprite preview generation (`pdftoppm`).
- **`imagemagick`** – Used for PSD format transcoding and color profile conversion to sRGB.
- **`ripgrep`** – Required by the AI agent sandbox (`anthropic-experimental/sandbox-runtime`).
Install the required packages with Homebrew:
```bash
brew install ffmpeg poppler imagemagick ripgrep
```
---
#### Windows (alpha)
> [!WARNING]
> Windows support in anthropic-experimental/sandbox-runtime is **alpha and the design is in flux**.
>
> The current implementation provides:
> - Network egress filtering via the **Windows Filtering Platform (WFP)**.
> - File read/write denial via **ACL stamping**.
>
> However, it is **not a security boundary against a deliberately adversarial sandboxed process**.
>
> The sandbox implementation will be substantially revised in a future release, where sandboxed processes will run under a dedicated sandbox user account.
Shumai requires:
- **`ffmpeg`** – Used for media transcoding and metadata extraction.
- **`imagemagick`** – Used for PSD format transcoding and color profile conversion to sRGB.
Install required packages using your preferred package manager:
**winget**
```powershell
winget install Gyan.FFmpeg ImageMagick.ImageMagick
```
**Chocolatey**
```powershell
choco install ffmpeg imagemagick
```
#### Step 4: Install Shumai globally
+115 -36
View File
@@ -17,25 +17,26 @@
## 功能特性
- **兼容 S3 与本地存储**:使用本地文件系统或任何兼容 S3 的云存储(AWS S3、Cloudflare R2、阿里云 OSS 等)安全地存储和提供您的创意资产。
- **逐帧批注与评论**:直接在视频和图像资产上使用特定帧的绘图工具和带时间戳的评论,提供精确的反馈。
- **安全分享与合集**:创建安全的公开分享链接和媒体合集,以便与客户和团队进行协作。
- **细粒度访问控制**:使用团队级别和项目级别的基于角色的访问控制来管理访问权限。
- **基于 Temporal 的分布式转码**:将占用大量资源的视频转码任务部署到 Temporal Worker上。
- **自定义资产元数据**:为媒体文件添加可搜索/编辑的自定义元数据字段。
- **兼容 S3 与本地存储**:支持使用本地文件系统或任意兼容 S3 协议的对象存储(AWS S3、Cloudflare R2、MinIO 等)安全存储和访问你的创意资产。
- **逐帧标注与评论**:可直接在图片和视频上进行逐帧绘制标注,并结合时间戳评论,实现精准反馈与协作。
- **安全分享与媒体合集**:支持创建安全的公开分享链接和精心组织的媒体合集,方便与客户及团队成员协同沟通。
- **细粒度权限控制**:基于团队和项目两级角色权限(RBAC),灵活管理工作区访问权限。
- **基于 Temporal 的分布式转码**:将高负载的视频转码任务交由 Temporal 调度的后台 Worker 集群处理,提升系统扩展性与处理效率。
- **自定义资产元数据**:支持定义和扩展动态元数据字段,灵活适配你的制作流程。
#### Shumai 智能体 (Agent)
#### Shumai Agent
- **AI Agent 协作**:通过评论区和 AI 智能体进行交流。
- **自定义技能**:通过添加自定义技能来扩展智能体的能力。
- **隔离的沙箱执行**:在沙箱环境中安全地运行智能体提交的脚本。
- **AI 元数据自动填充**:使用AI为资产自动生填充元数据字段。
- **语义搜索**:基于 Gemini Embedding 2 的图片/视频内容搜索。
- **协作式 AI 对话**:在文件评论区通过 `@mention` 即可唤起具备上下文感知能力的 AI Agent,参与讨论与协作。
- **自定义技能与工具**:支持为 Agent 注册自定义脚本、工具和自动化技能,扩展其能力边界。
- **隔离沙箱执行**:Agent 提交的脚本会在安全隔离的沙箱环境中运行,保障执行安全。
- **AI 自动填充元数据**:为新上传的资产自动生成并填写自定义元数据。
- **语义搜索**:基于向量进行搜索(Gemini Embedding 2),支持按视觉内容或语义概念搜索资产。
> [!NOTE]
> **媒体处理与存储说明:** 在上传媒体文件时,Shumai 会始终完整保留并存储您的原始文件。对于所有媒体类型(图片与视频),Shumai 都会统一生成一张 480p 的 WebP 预览图,用作文件列表页预览。此外:
> * **图片文件:** Shumai 会额外生成一个与原图分辨率完全一致的 WebP 优化版本。
> * **视频文件:** 系统会根据您的转码设置生成至少一个预览版本mp4,同时还会生成一张雪碧图(Sprite)和一个轻量级预览视频(上限 1600 帧),以实现流畅的文件列表页面预览和文件查看页面时间轴快速预览。
> **媒体处理与存储机制:** 上传媒体文件后,Shumai 始终会保留并存储原始文件。请注意,系统中显示的文件大小为原始文件的大小,不包含转码后生成的代理文件大小。无论图片还是视频,都会额外生成一张 480p 的 WebP 封面图,作为文件列表中的预览图。此外:
>
> * **图片:** 会生成一份与原图分辨率完全一致的高质量 WebP 转码版本,在保证画质的同时提升浏览性能。
> * **视频:** 会根据转码配置生成至少一个代理视频,同时生成雪碧图(Sprite)和一段最长 1600 帧的低分辨率转码视频,用于实现快速拖动时间轴和即时视频预览。
---
@@ -62,15 +63,26 @@
3. 配置环境变量(可选):
* `SHUMAI_SERVER_PORT` 用于控制 Shumai 服务监听的端口,默认值为 `3000`。
* 默认情况下,该 Docker Compose 部署会自动启用集成的本地存储。若想切换至外部 S3 兼容存储,请将 `AWS_ENDPOINT_URL_S3` 设置为对应云厂商的 Endpoint URL,并根据需要配置相应的 AWS 凭证和存储桶(Bucket)参数:
```
AWS_ENDPOINT_URL_S3: S3_ENDPOINT_URL
S3_REGION: S3_REGION
S3_BUCKET: BUCKET_NAME
S3_ACCESS_KEY_ID: ID
S3_SECRET_ACCESS_KEY: KEY
* 默认情况下,该 Docker Compose 部署会自动启用集成的本地存储。若想切换至外部 S3 兼容存储,请在 `environment` 中配置 S3 提供商的环境变量。以下是 AWS S3 和 Cloudflare R2 的配置示例:
```
**AWS S3 示例:**
```yaml
STORAGE_BACKEND: s3
S3_BUCKET: your-bucket-name
S3_ACCESS_KEY_ID: your-access-key-id
S3_SECRET_ACCESS_KEY: your-secret-access-key
AWS_ENDPOINT_URL_S3: https://s3.us-east-1.amazonaws.com
```
**Cloudflare R2 示例:**
```yaml
STORAGE_BACKEND: s3
S3_BUCKET: your-bucket-name
S3_REGION: auto
S3_ACCESS_KEY_ID: your-access-key-id
S3_SECRET_ACCESS_KEY: your-secret-access-key
AWS_ENDPOINT_URL_S3: https://<account-id>.r2.cloudflarestorage.com
```
* 默认情况下 `AWS_ENDPOINT_URL_S3` 指向 `http://localhost:{SHUMAI_SERVER_PORT}`。如果您使用本地存储,但希望在自定义的主机名或端口上公开访问 Shumai,请务必将 `AWS_ENDPOINT_URL_S3` 修改为浏览器访问该服务时所使用的外部 URL。
@@ -120,26 +132,93 @@ mkdir shumai && cd shumai
```
#### 第三步:安装平台特定系统依赖
#### 步骤 3:安装平台依赖
在安装 Shumai 之前,请确保您的宿主机上已经安装了以下系统依赖项:
#### Linux
| 依赖包 | 功能描述 | Ubuntu/Debian | Fedora | Arch | macOS |
| --- | --- | --- | --- | --- | --- |
| **ffmpeg** | 媒体转码及元数据提取 | `sudo apt install -y ffmpeg` | `sudo dnf install -y ffmpeg` | `sudo pacman -S --noconfirm ffmpeg` | `brew install ffmpeg` |
| **bubblewrap** | 用于安全执行 AI 助手脚本的隔离沙箱环境 | `sudo apt install -y bubblewrap` | `sudo dnf install -y bubblewrap` | `sudo pacman -S --noconfirm bubblewrap` | *无需安装* |
| **socat** | 用于沙箱网络桥接的双向套接字中继 | `sudo apt install -y socat` | `sudo dnf install -y socat` | `sudo pacman -S --noconfirm socat` | `brew install socat` |
| **ripgrep** | 用于工作区安全策略的高效文本搜索工具 | `sudo apt install -y ripgrep` | `sudo dnf install -y ripgrep` | `sudo pacman -S --noconfirm ripgrep` | `brew install ripgrep` |
在 Linux 上运行 Shumai 需要安装以下系统依赖:
- **`ffmpeg`** —— 用于媒体转码和元数据提取。
- **`poppler`** (`poppler-utils`) —— 用于 PDF 页面图像提取和 PDF 雪碧图预览生成 (`pdftoppm`)。
- **`imagemagick`** —— 用于 PSD 格式图像转码和 sRGB 色彩空间转换。
- **`bubblewrap`**、**`socat`** 和 **`ripgrep`** —— AI Agent 沙箱(`anthropic-experimental/sandbox-runtime`)所需,用于进程隔离、网络通信和工作区搜索。
可以使用以下命令一次性安装所有依赖:
> Ubuntu/Debian
```bash
sudo apt install -y ffmpeg poppler-utils imagemagick bubblewrap socat ripgrep
```
> Fedora
```bash
sudo dnf install -y ffmpeg poppler-utils ImageMagick bubblewrap socat ripgrep
```
> Arch Linux
```bash
sudo pacman -S --noconfirm ffmpeg poppler imagemagick bubblewrap socat ripgrep
```
> [!NOTE]
> **Ubuntu 24.04+ 用户请注意:** 这些系统版本默认限制了非特权用户命名空间(unprivileged user namespaces)。为了让 `bubblewrap` 和沙箱隔离层正常工作,需要关闭此限制:
> **Ubuntu 24.04+** 默认限制非特权用户创建 User Namespace,这会导致 `anthropic-experimental/sandbox-runtime` 无法使用 `bubblewrap` 提供沙箱隔离。
>
> 如需临时启用 User Namespace,请执行:
>
> ```bash
> sudo sysctl -w kernel.apparmor_restrict_unprivileged_userns=0
>
> ```
>
>
> 或者,您也可以专门配置一个 AppArmor 配置文件,为相关二进制文件授予用户命名空间创建(`userns`)权限。
>
> 或者,配置 AppArmor Profile,为相关可执行文件授予所需的 `userns` 权限。
---
#### macOS
在 macOS 上运行 Shumai 需要:
- **`ffmpeg`** —— 用于媒体转码和元数据提取。
- **`poppler`** —— 用于 PDF 页面图像提取和 PDF 雪碧图预览生成 (`pdftoppm`)。
- **`imagemagick`** —— 用于 PSD 格式图像转码和 sRGB 色彩空间转换。
- **`ripgrep`** —— AI Agent 沙箱(`anthropic-experimental/sandbox-runtime`)所需。
使用 Homebrew 安装所需依赖:
```bash
brew install ffmpeg poppler imagemagick ripgrep
```
---
#### Windows(Alpha)
> [!WARNING]
> anthropic-experimental/sandbox-runtime 的 Windows 支持目前仍处于 **Alpha** 阶段,整体设计仍在持续调整中。
>
> 当前实现提供了以下能力:
> - 通过 **Windows Filtering Platform (WFP)** 实现网络出口(egress)过滤。
> - 通过 **ACL Stamping** 实现文件读写权限限制。
>
> **但这并不能作为针对恶意沙箱进程的安全边界(security boundary)**。
>
> 未来版本将对沙箱机制进行重大调整,届时沙箱进程将运行在独立的沙箱用户账户下。
在 Windows 上运行 Shumai 需要:
- **`ffmpeg`** —— 用于媒体转码和元数据提取。
- **`imagemagick`** —— 用于 PSD 格式图像转码和 sRGB 色彩空间转换。
使用你偏好的包管理器安装所需依赖:
**winget**
```powershell
winget install Gyan.FFmpeg ImageMagick.ImageMagick
```
**Chocolatey**
```powershell
choco install ffmpeg imagemagick
```
#### 第四步:全局安装 Shumai
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "@shumai/agent-app",
"version": "0.1.6",
"version": "0.2.2",
"private": true,
"type": "module",
"bin": {
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "@shumai/cli",
"version": "0.1.6",
"version": "0.2.2",
"private": true,
"type": "module",
"bin": {
+2
View File
@@ -33,6 +33,8 @@ export function getMediaType(filename: string): string {
'.wav': 'audio/wav',
'.ogg': 'audio/ogg',
'.txt': 'text/plain',
'.md': 'text/markdown',
'.markdown': 'text/markdown',
'.json': 'application/json',
'.pdf': 'application/pdf',
'.zip': 'application/zip',
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "@shumai/transcode-app",
"version": "0.1.6",
"version": "0.2.2",
"private": true,
"type": "module",
"bin": {
+56
View File
@@ -0,0 +1,56 @@
import { test as base } from '@playwright/test'
import { PrismaClient } from '../../../packages/db/src/generated/prisma/client'
import { PrismaPg } from '@prisma/adapter-pg'
import { Pool } from 'pg'
import * as fs from 'fs'
import * as path from 'path'
import { fileURLToPath } from 'url'
const currentDir = path.dirname(fileURLToPath(import.meta.url))
const envPath = path.resolve(currentDir, '.env.e2e')
// Load E2E environment variables if present
if (fs.existsSync(envPath)) {
const content = fs.readFileSync(envPath, 'utf8')
content.split('\n').forEach((line) => {
const parts = line.split('=')
if (parts.length >= 2) {
const key = parts[0].trim()
const value = parts.slice(1).join('=').trim()
process.env[key] = value
}
})
}
export const test = base.extend<{ prisma: PrismaClient }>({
prisma: [
// eslint-disable-next-line no-empty-pattern
async ({}, use) => {
const connectionString = process.env.DATABASE_URL
const pool = new Pool({ connectionString })
const adapter = new PrismaPg(pool)
const prisma = new PrismaClient({ adapter })
// Reset database state before the test runs
await prisma.$executeRawUnsafe(`
DO $$ DECLARE
r RECORD;
BEGIN
FOR r IN (SELECT tablename FROM pg_tables WHERE schemaname = 'public' AND tablename <> '_prisma_migrations') LOOP
EXECUTE 'TRUNCATE TABLE ' || quote_ident(r.tablename) || ' CASCADE;';
END LOOP;
END $$;
`)
// Provide the clean prisma instance to the test
await use(prisma)
// Disconnect and clean up resources after the test finishes
await prisma.$disconnect()
await pool.end()
},
{ auto: true },
],
})
export { expect } from '@playwright/test'
+36
View File
@@ -0,0 +1,36 @@
import { teardown as dbTeardown } from '../../../packages/db/src/test-global-setup'
import * as fs from 'fs'
import * as path from 'path'
import { fileURLToPath } from 'url'
const currentDir = path.dirname(fileURLToPath(import.meta.url))
export default async function globalTeardown() {
const envPath = path.resolve(currentDir, '.env.e2e')
if (fs.existsSync(envPath)) {
const content = fs.readFileSync(envPath, 'utf8')
let bucketName: string | undefined
content.split('\n').forEach((line) => {
const parts = line.split('=')
if (parts.length >= 2 && parts[0].trim() === 'S3_BUCKET') {
bucketName = parts.slice(1).join('=').trim()
}
})
if (bucketName) {
const bucketDir = path.resolve('data', bucketName)
if (fs.existsSync(bucketDir)) {
console.log(`Cleaning up E2E local storage bucket: ${bucketDir}`)
fs.rmSync(bucketDir, { recursive: true, force: true })
}
}
// Delete the temp .env.e2e file
fs.unlinkSync(envPath)
}
// Stop pgvector container
await dbTeardown()
console.log('Global teardown complete.')
}
+35
View File
@@ -0,0 +1,35 @@
import { globalTestSetup } from '../../../packages/db/src/test-global-setup'
import { randomBytes } from 'crypto'
import * as fs from 'fs'
import * as path from 'path'
import { fileURLToPath } from 'url'
const currentDir = path.dirname(fileURLToPath(import.meta.url))
async function start() {
console.log('Starting E2E test database container...')
await globalTestSetup()
// Generate a random S3 bucket name
const randomSuffix = randomBytes(4).toString('hex')
const bucketName = `shumai-e2e-${randomSuffix}`
process.env.S3_BUCKET = bucketName
process.env.SHUMAI_SERVER_PORT = '5200'
process.env.BETTER_AUTH_URL = 'http://localhost:5200'
process.env.NODE_ENV = 'development'
process.env.SHUMAI_E2E = 'true'
// Write variables to .env.e2e file so the teardown can load them
const envContent = [`DATABASE_URL=${process.env.DATABASE_URL}`, `S3_BUCKET=${bucketName}`].join(
'\n',
)
fs.writeFileSync(path.resolve(currentDir, '.env.e2e'), envContent)
console.log(`E2E env file written. Starting Web App...`)
// Now run the web app
await import('../src/index.ts')
}
start().catch(console.error)
+75
View File
@@ -0,0 +1,75 @@
import { expect, test } from './fixtures'
import * as fs from 'fs'
import * as path from 'path'
import { fileURLToPath } from 'url'
const currentDir = path.dirname(fileURLToPath(import.meta.url))
const tempImgPath = path.resolve(currentDir, 'temp-cover.png')
test.beforeAll(async () => {
// Create a 1x1 transparent PNG file
const pngBase64 =
'iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII='
fs.writeFileSync(tempImgPath, Buffer.from(pngBase64, 'base64'))
})
test.afterAll(async () => {
if (fs.existsSync(tempImgPath)) {
fs.unlinkSync(tempImgPath)
}
})
test('should sign up and create a project with a cover image', async ({ page }) => {
const email = `test-${Date.now()}@shumai.local`
const password = 'Password123!'
// 1. Sign Up Flow
await page.goto('/signup')
await expect(page).toHaveURL(/\/signup/)
// Fill credentials
await page.fill('#email', email)
await page.fill('#password', password)
// Submit sign up
await page.click('button[type="submit"]')
// Wait for the automatic redirect to the team page (since only one team exists)
await expect(page).toHaveURL(/\/teams\/[^/]+/)
// 2. Create Project Flow
// Click the "Create Project" button on the dashboard
await page.click('button:has-text("Create Project")')
const dialog = page.locator('[role="dialog"]')
await expect(dialog).toBeVisible()
// Fill project name
const projectName = `My Project ${Date.now()}`
await dialog.locator('#name').fill(projectName)
// Set cover image
await dialog.locator('input[type="file"]').setInputFiles(tempImgPath)
// Wait for image upload completion and preview
await expect(dialog.locator('img[alt="Cover Preview"]')).toBeVisible({ timeout: 15_000 })
// Click submit in dialog
await dialog.locator('button[type="submit"]').click()
// Wait for navigation to the project page
await expect(page).toHaveURL(/\/projects\/[^/]+/)
// Verify project name appears in the project view
await expect(page.locator(`text=${projectName}`).first()).toBeVisible()
// 3. Verification on Dashboard
// Navigate back to the dashboard via sidebar
await page.click('button[aria-label="Dashboard"]')
// Wait for dashboard URL
await expect(page).toHaveURL(/\/teams\/[^/]+/)
// Verify the project card is visible and displays the cover image with project name alt text
await expect(page.locator(`img[alt="${projectName}"]`)).toBeVisible()
})
+6 -1
View File
@@ -1,6 +1,6 @@
{
"name": "@shumai/web-app",
"version": "0.1.6",
"version": "0.2.2",
"private": true,
"type": "module",
"dependencies": {
@@ -13,5 +13,10 @@
"@shumai/transcode": "workspace:*",
"@shumai/webui": "workspace:*",
"hono": "4.12.23"
},
"devDependencies": {
"@prisma/adapter-pg": "7.8.0",
"pg": "8.21.0",
"@types/pg": "8.20.0"
}
}
+41 -6
View File
@@ -1,5 +1,6 @@
import { loadEnvConfig } from '@shumai/core/src/env-loader'
import { join } from 'node:path'
import { readdirSync } from 'node:fs'
import { dirname, join } from 'node:path'
loadEnvConfig(process.cwd(), process.env.NODE_ENV !== 'production')
import index from '@shumai/webui/index.html'
@@ -84,12 +85,38 @@ async function run() {
routes[routePath] = () => new Response(Bun.file(filePath), { headers: file.headers })
}
const sampleFile = bundle.files[0]?.path || bundle.index || ''
const bundleDir = join(import.meta.dir, dirname(sampleFile))
try {
const dirFiles = readdirSync(bundleDir)
const pdfWorkerFile = dirFiles.find((f) => f.startsWith('pdf.worker'))
if (pdfWorkerFile) {
const routePath = '/' + pdfWorkerFile
const filePath = join(bundleDir, pdfWorkerFile)
const bunFile = Bun.file(filePath)
routes[routePath] = () =>
new Response(bunFile, {
headers: {
'content-type': 'text/javascript;charset=utf-8',
etag: `"${bunFile.size}-${bunFile.lastModified}"`,
},
})
}
} catch (err) {
console.warn('Failed to register pdf.worker route:', err)
}
const mainHtmlPath = join(import.meta.dir, bundle.index)
const mainHtmlFile = Bun.file(mainHtmlPath)
const mainHtmlHeaders = bundle.files.find((f) => f.path === bundle.index)?.headers || {
'content-type': 'text/html;charset=utf-8',
}
const serveMainHtml = () => new Response(mainHtmlFile, { headers: mainHtmlHeaders })
const raw = await mainHtmlFile.text()
const mainHtmlText = raw
.replaceAll('href="./', 'href="/')
.replaceAll('src="./', 'src="/')
.replaceAll('content="./', 'content="/')
const serveMainHtml = () => new Response(mainHtmlText, { headers: mainHtmlHeaders })
routes['/'] = serveMainHtml
routes['/*'] = serveMainHtml
@@ -109,6 +136,7 @@ async function run() {
const server = Bun.serve({
port,
idleTimeout: 120,
maxRequestBodySize: process.env.MAX_REQUEST_BODY_SIZE
? parseInt(process.env.MAX_REQUEST_BODY_SIZE)
: 1024 * 1024 * 1024 * 10, // Default 10GB
@@ -130,7 +158,14 @@ async function run() {
process.on('SIGTERM', shutdown)
}
handleDaemonCommands('shumai', run).catch((err) => {
console.error(err)
process.exit(1)
})
if (process.env.SHUMAI_E2E === 'true') {
run().catch((err) => {
console.error(err)
process.exit(1)
})
} else {
handleDaemonCommands('shumai', run).catch((err) => {
console.error(err)
process.exit(1)
})
}
+3 -2
View File
@@ -19,7 +19,8 @@ await Bun.build({
root: '.',
target: 'bun',
outdir: outdir,
minify: false,
minify: true,
splitting: true,
naming: '[name].[ext]',
plugins: [
temporalWorkflow({
@@ -33,6 +34,7 @@ await Bun.build({
'@temporalio/client',
'@temporalio/worker',
'@temporalio/workflow',
'pdfkit',
],
// compile: {
// outfile: 'Shumai',
@@ -44,5 +46,4 @@ await Bun.build({
define: {
'process.env.NODE_ENV': JSON.stringify('production'),
},
publicPath: '/',
})
+116 -20
View File
@@ -4,10 +4,6 @@
"workspaces": {
"": {
"name": "shumai",
"dependencies": {
"@tanstack/react-virtual": "3.14.2",
"jittered-fractional-indexing": "1.0.1",
},
"devDependencies": {
"@eslint/js": "10.0.1",
"@inlang/paraglide-js": "2.20.0",
@@ -37,7 +33,7 @@
},
"apps/agent": {
"name": "@shumai/agent-app",
"version": "0.1.5",
"version": "0.2.0",
"bin": {
"shumai-agent": "./src/index.ts",
},
@@ -49,7 +45,7 @@
},
"apps/cli": {
"name": "@shumai/cli",
"version": "0.1.5",
"version": "0.2.0",
"bin": {
"shumai-cli": "./src/index.ts",
},
@@ -64,7 +60,7 @@
},
"apps/transcode": {
"name": "@shumai/transcode-app",
"version": "0.1.5",
"version": "0.2.0",
"bin": {
"shumai-transcode": "./src/index.ts",
},
@@ -76,7 +72,7 @@
},
"apps/web": {
"name": "@shumai/web-app",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@shumai/agent": "workspace:*",
"@shumai/api": "workspace:*",
@@ -88,14 +84,19 @@
"@shumai/workflow-core": "workspace:*",
"hono": "4.12.23",
},
"devDependencies": {
"@prisma/adapter-pg": "7.8.0",
"@types/pg": "8.20.0",
"pg": "8.21.0",
},
},
"packages/agent": {
"name": "@shumai/agent",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@anthropic-ai/sandbox-runtime": "0.0.52",
"@earendil-works/pi-agent-core": "0.78.0",
"@earendil-works/pi-ai": "0.75.4",
"@earendil-works/pi-ai": "0.80.7",
"@google/genai": "1.52.0",
"@shumai/core": "workspace:*",
"@shumai/db": "workspace:*",
@@ -106,6 +107,7 @@
"@temporalio/worker": "1.17.2",
"@temporalio/workflow": "1.17.2",
"adm-zip": "0.5.17",
"jittered-fractional-indexing": "1.0.1",
"ulid": "3.0.2",
"zod": "4.4.3",
},
@@ -115,7 +117,7 @@
},
"packages/api": {
"name": "@shumai/api",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@hono/zod-validator": "0.7.6",
"@shumai/core": "workspace:*",
@@ -129,19 +131,22 @@
},
"packages/core": {
"name": "@shumai/core",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@earendil-works/pi-agent-core": "0.78.0",
"@shumai/db": "workspace:*",
"@shumai/dtos": "workspace:*",
"@sinclair/typebox": "0.34.49",
"adm-zip": "0.5.17",
"better-auth": "1.6.11",
"csv-parse": "7.0.1",
"date-fns": "4.4.0",
"dotenv": "16.6.1",
"dotenv-expand": "13.0.0",
"extract-zip": "2.0.1",
"hono": "4.12.23",
"jittered-fractional-indexing": "1.0.1",
"pdfkit": "0.19.1",
"pino": "10.3.1",
"sharp": "0.34.5",
"tweetnacl": "1.0.3",
@@ -152,11 +157,12 @@
"devDependencies": {
"@types/adm-zip": "0.5.8",
"@types/extract-zip": "2.0.3",
"@types/pdfkit": "0.17.6",
},
},
"packages/db": {
"name": "@shumai/db",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@earendil-works/pi-agent-core": "0.78.0",
"@prisma/adapter-pg": "7.8.0",
@@ -172,18 +178,33 @@
},
"packages/dtos": {
"name": "@shumai/dtos",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@earendil-works/pi-agent-core": "0.78.0",
"zod": "4.4.3",
},
},
"packages/e2e": {
"name": "@shumai/e2e",
"version": "0.2.0",
"dependencies": {
"@earendil-works/pi-agent-core": "0.78.0",
"@google/genai": "1.52.0",
"@shumai/agent": "workspace:*",
"@shumai/core": "workspace:*",
"@shumai/db": "workspace:*",
"@shumai/dtos": "workspace:*",
"@shumai/transcode": "workspace:*",
"@shumai/workflow-core": "workspace:*",
},
},
"packages/tableprinter": {
"name": "@shumai/tableprinter",
"version": "0.1.5",
"version": "0.2.0",
},
"packages/transcode": {
"name": "@shumai/transcode",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@shumai/core": "workspace:*",
"@shumai/db": "workspace:*",
@@ -197,7 +218,7 @@
},
"packages/webui": {
"name": "@shumai/webui",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@base-ui/react": "1.5.0",
"@dnd-kit/abstract": "0.4.0",
@@ -259,6 +280,7 @@
"lexical": "0.42.0",
"lucide-react": "1.17.0",
"next-themes": "0.4.6",
"pdfjs-dist": "6.1.200",
"radix-ui": "1.4.3",
"react": "19.2.0",
"react-day-picker": "10.0.1",
@@ -294,7 +316,7 @@
},
"packages/workflow-core": {
"name": "@shumai/workflow-core",
"version": "0.1.5",
"version": "0.2.0",
"dependencies": {
"@shumai/core": "workspace:*",
"@shumai/db": "workspace:*",
@@ -443,7 +465,7 @@
"@earendil-works/pi-agent-core": ["@earendil-works/pi-agent-core@0.78.0", "", { "dependencies": { "@earendil-works/pi-ai": "^0.78.0", "ignore": "7.0.5", "typebox": "1.1.38", "yaml": "2.9.0" } }, "sha512-xhWd59Qzd8yO88gYQw2S4dEQstJJEiUtxRP01//YzVJ61jCtUASMfcyAmYhgGYR4Onp7GmwEAbBBGOiV6Iwk9g=="],
"@earendil-works/pi-ai": ["@earendil-works/pi-ai@0.75.4", "", { "dependencies": { "@anthropic-ai/sdk": "0.91.1", "@aws-sdk/client-bedrock-runtime": "3.1048.0", "@google/genai": "1.52.0", "@mistralai/mistralai": "2.2.1", "http-proxy-agent": "7.0.2", "https-proxy-agent": "7.0.6", "openai": "6.26.0", "partial-json": "0.1.7", "typebox": "1.1.38" }, "bin": { "pi-ai": "dist/cli.js" } }, "sha512-m/w8Hh3vQ0rAycwJiJWdzkypkn4295f4eq/966lDRy8aX5sk6bgYXH8TQmL16TO7Uwc7MbJG0QoyFHgX8RqXUQ=="],
"@earendil-works/pi-ai": ["@earendil-works/pi-ai@0.80.7", "", { "dependencies": { "@anthropic-ai/sdk": "0.91.1", "@aws-sdk/client-bedrock-runtime": "3.1048.0", "@google/genai": "1.52.0", "@mistralai/mistralai": "2.2.6", "@opentelemetry/api": "1.9.0", "@smithy/node-http-handler": "4.7.3", "http-proxy-agent": "7.0.2", "https-proxy-agent": "7.0.6", "openai": "6.26.0", "partial-json": "0.1.7", "typebox": "1.1.38" }, "bin": { "pi-ai": "dist/cli.js" } }, "sha512-8RLKLwe5TFM9kKFMNu/lTzveduq4GxZbnlG6ba8FAhLeb5wJP4zbj1eBumKBRvggpFQnW5R/Vo2a8zTlHsV9SQ=="],
"@electric-sql/pglite": ["@electric-sql/pglite@0.4.1", "", {}, "sha512-mZ9NzzUSYPOCnxHH1oAHPRzoMFJHY472raDKwXl/+6oPbpdJ7g8LsCN4FSaIIfkiCKHhb3iF/Zqo3NYxaIhU7Q=="],
@@ -709,7 +731,31 @@
"@lix-js/server-protocol-schema": ["@lix-js/server-protocol-schema@0.1.1", "", {}, "sha512-jBeALB6prAbtr5q4vTuxnRZZv1M2rKe8iNqRQhFJ4Tv7150unEa0vKyz0hs8Gl3fUGsWaNJBh3J8++fpbrpRBQ=="],
"@mistralai/mistralai": ["@mistralai/mistralai@2.2.1", "", { "dependencies": { "ws": "^8.18.0", "zod": "^3.25.0 || ^4.0.0", "zod-to-json-schema": "^3.25.0" } }, "sha512-uKU8CZmL2RzYKmplsU01hii4p3pe4HqJefpWNRWXm1Tcm0Sm4xXfwSLIy4k7ZCPlbETCGcp69E7hZs+WOJ5itQ=="],
"@mistralai/mistralai": ["@mistralai/mistralai@2.2.6", "", { "dependencies": { "@opentelemetry/semantic-conventions": "^1.40.0", "ws": "^8.18.0", "zod": "^3.25.0 || ^4.0.0", "zod-to-json-schema": "^3.25.0" }, "peerDependencies": { "@opentelemetry/api": "^1.9.0" }, "optionalPeers": ["@opentelemetry/api"] }, "sha512-W8pX7zHxjJvMIpw8JMxeJEleapXX0Q9NPszdNzqkM3MIEoIGPObdodujj+WHteXEvGfaP/AMwlNyRfEzSY6dQQ=="],
"@napi-rs/canvas": ["@napi-rs/canvas@1.0.2", "", { "optionalDependencies": { "@napi-rs/canvas-android-arm64": "1.0.2", "@napi-rs/canvas-darwin-arm64": "1.0.2", "@napi-rs/canvas-darwin-x64": "1.0.2", "@napi-rs/canvas-linux-arm-gnueabihf": "1.0.2", "@napi-rs/canvas-linux-arm64-gnu": "1.0.2", "@napi-rs/canvas-linux-arm64-musl": "1.0.2", "@napi-rs/canvas-linux-riscv64-gnu": "1.0.2", "@napi-rs/canvas-linux-x64-gnu": "1.0.2", "@napi-rs/canvas-linux-x64-musl": "1.0.2", "@napi-rs/canvas-win32-arm64-msvc": "1.0.2", "@napi-rs/canvas-win32-x64-msvc": "1.0.2" } }, "sha512-EYEqlMYaCbpZDz+IgDH5xp9MTd3ui4dmGqbQYryhMLnSRxrhHKq5KQWHHKxFUcEP4Hp8/BWgvqXocX4j7iSbOQ=="],
"@napi-rs/canvas-android-arm64": ["@napi-rs/canvas-android-arm64@1.0.2", "", { "os": "android", "cpu": "arm64" }, "sha512-IMXKVQod0ol4vt3gmClUfXz4JAgHYESGPCUqmH3lQxBoL0K/2greJaQE1HVBVxWWFKfLc4OLZVdxg7kXVyXv+g=="],
"@napi-rs/canvas-darwin-arm64": ["@napi-rs/canvas-darwin-arm64@1.0.2", "", { "os": "darwin", "cpu": "arm64" }, "sha512-Sc8tPi6cF+5lqOzCCKFALJHhDiRwyMzTPYm3bbhdXsOunU0lQO5f05ucyOzN2r55I23Hg5bsjH63uSCvWp3EgQ=="],
"@napi-rs/canvas-darwin-x64": ["@napi-rs/canvas-darwin-x64@1.0.2", "", { "os": "darwin", "cpu": "x64" }, "sha512-niDXZ9LhKB1zLrUdYB64RHQFDGz9rr0eGx061qtJJU3U20EMMIx28ADF5fVYbhtOgkWQrBjFicfaye1yM0U62A=="],
"@napi-rs/canvas-linux-arm-gnueabihf": ["@napi-rs/canvas-linux-arm-gnueabihf@1.0.2", "", { "os": "linux", "cpu": "arm" }, "sha512-sgatQL9JxGRH/Amzcvu0P3t8Am3duou74CisfuJ41Dwt8cWy723z/9KZ8LlgmxfypEwEZxSTNFJtU8d281lmhQ=="],
"@napi-rs/canvas-linux-arm64-gnu": ["@napi-rs/canvas-linux-arm64-gnu@1.0.2", "", { "os": "linux", "cpu": "arm64" }, "sha512-dgKuX0peF3xwY6ZF5QxGS4wbfDqpoFAJYXiLSp+guZKARQUKMkRqZSDrXKj7nfrec3UCMzC0PFCPte0ES98AiA=="],
"@napi-rs/canvas-linux-arm64-musl": ["@napi-rs/canvas-linux-arm64-musl@1.0.2", "", { "os": "linux", "cpu": "arm64" }, "sha512-qwROoDIC9upfvDoRLuPn2aNg9CGW1x0Ygr4k2Or+8paA9d0qBLwk87U+g8KQpoOviKoPoiwl97kvBYuYD7qZoA=="],
"@napi-rs/canvas-linux-riscv64-gnu": ["@napi-rs/canvas-linux-riscv64-gnu@1.0.2", "", { "os": "linux", "cpu": "none" }, "sha512-fXRjnPihdnbO6qy1QQOgxAonb68A0TCEG7rj1x7v7rxNElsE8EVIKIEUTvyDtU+sthYSbX+8e7g3oZiLGnOmxw=="],
"@napi-rs/canvas-linux-x64-gnu": ["@napi-rs/canvas-linux-x64-gnu@1.0.2", "", { "os": "linux", "cpu": "x64" }, "sha512-nPR97DXhbWIAy7yazF3jc06kEPMqYMLmPzFOVNlwKPfIoSChnI+x7dc0hTLaihz3jxrjL6j4BbA7earxfx4X3g=="],
"@napi-rs/canvas-linux-x64-musl": ["@napi-rs/canvas-linux-x64-musl@1.0.2", "", { "os": "linux", "cpu": "x64" }, "sha512-l7zZY5+jL5qnBZtDz7CoBtY6p7EkHu422g/0zWwrOrzIwWyWxZFRfZZORY1UG7YApymPLx+UbOkN206xXn/c1Q=="],
"@napi-rs/canvas-win32-arm64-msvc": ["@napi-rs/canvas-win32-arm64-msvc@1.0.2", "", { "os": "win32", "cpu": "arm64" }, "sha512-yE0koHCFF4PIbMc2o2SEALhnipz7WBISh5glLvQiomtIoCcW0np3H4Lw93ceJAfJttTTeIIWFbwH84F7EVzjMQ=="],
"@napi-rs/canvas-win32-x64-msvc": ["@napi-rs/canvas-win32-x64-msvc@1.0.2", "", { "os": "win32", "cpu": "x64" }, "sha512-okU8/t2foV6C31n0GtvEMbfD5rOFc70+/6xUNME9Guld29sgSOIGUEDScAWFlcP3k5TYQRl9TNkwJEEjh15w8A=="],
"@noble/ciphers": ["@noble/ciphers@2.2.0", "", {}, "sha512-Z6pjIZ/8IJcCGzb2S/0Px5J81yij85xASuk1teLNeg75bfT07MV3a/O2Mtn1I2se43k3lkVEcFaR10N4cgQcZA=="],
@@ -717,6 +763,8 @@
"@nodable/entities": ["@nodable/entities@2.1.1", "", {}, "sha512-Pig3HxDIoMgjdEH8OCf/dkcTmLFjJRjWuq8jSnklu284/TKOPibSRERmOykiwmyXTtv61mP+44f3GMx0tLAyjg=="],
"@opentelemetry/api": ["@opentelemetry/api@1.9.0", "", {}, "sha512-3giAOQvZiH5F9bMlMiv8+GSPMeqg0dbaeo58/0SlA9sxSqZhnUtxzX9/2FzyhS9sWQf5S0GJE0AKBrFqjpeYcg=="],
"@opentelemetry/semantic-conventions": ["@opentelemetry/semantic-conventions@1.41.1", "", {}, "sha512-/UhIkaZgPutTFmQ7RnIJGgDXZmtEJ7Dvi86xNTFWcnRxVRNk/aotsqDJYeEvDP+FSMB2SdW+pQzNMcWP0rwuNA=="],
"@oven/bun-darwin-aarch64": ["@oven/bun-darwin-aarch64@1.3.14", "", { "os": "darwin", "cpu": "arm64" }, "sha512-Omj20SuiHBOUjUBIyqtkNjSUIjOtEOJwmbix/ZyFH4BaQ6OZTaaRWIR4TjHVz0yadHgli6lLTiAh1uarnvD49A=="],
@@ -1003,6 +1051,8 @@
"@shumai/dtos": ["@shumai/dtos@workspace:packages/dtos"],
"@shumai/e2e": ["@shumai/e2e@workspace:packages/e2e"],
"@shumai/tableprinter": ["@shumai/tableprinter@workspace:packages/tableprinter"],
"@shumai/transcode": ["@shumai/transcode@workspace:packages/transcode"],
@@ -1069,6 +1119,8 @@
"@swc/counter": ["@swc/counter@0.1.3", "", {}, "sha512-e2BR4lsJkkRlKZ/qCHPw9ZaSxc0MVUd7gtbtaB7aMvHeJVYe8sOB8DBZkP2DtISHGSku9sCK6T6cnY0CtXrOCQ=="],
"@swc/helpers": ["@swc/helpers@0.5.23", "", { "dependencies": { "tslib": "^2.8.0" } }, "sha512-5lSsMOTXURePglDfvuAQUqkGek9Hg2kksOYay2m0+XR++b2NWYL/4sWyuvVBIs8oKnJaxkdi9whaL/sqN13afw=="],
"@swc/types": ["@swc/types@0.1.26", "", { "dependencies": { "@swc/counter": "^0.1.3" } }, "sha512-lyMwd7WGgG79RS7EERZV3T8wMdmPq3xwyg+1nmAM64kIhx5yl+juO2PYIHb7vTiPgPCj8LYjsNV2T5wiQHUEaw=="],
"@tailwindcss/node": ["@tailwindcss/node@4.3.0", "", { "dependencies": { "@jridgewell/remapping": "^2.3.5", "enhanced-resolve": "^5.21.0", "jiti": "^2.6.1", "lightningcss": "1.32.0", "magic-string": "^0.30.21", "source-map-js": "^1.2.1", "tailwindcss": "4.3.0" } }, "sha512-aFb4gUhFOgdh9AXo4IzBEOzBkkAxm9VigwDJnMIYv3lcfXCJVesNfbEaBl4BNgVRyid92AmdviqwBUBRKSeY3g=="],
@@ -1215,6 +1267,8 @@
"@types/node": ["@types/node@25.9.1", "", { "dependencies": { "undici-types": ">=7.24.0 <7.24.7" } }, "sha512-xfrlY7UD5rMJk3ZVJP8BNzS28J36YJg+xp+LPXV1TdWxr8uMH5A860QNxYDGQe/ylDSgjxE52Q9VnO7p75tJxg=="],
"@types/pdfkit": ["@types/pdfkit@0.17.6", "", { "dependencies": { "@types/node": "*" } }, "sha512-tIwzxk2uWKp0Cq9JIluQXJid77lYhF52EsIOwhsMF4iWLA6YneoBR1xVKYYdAysHuepUB0OX4tdwMiUDdGKmig=="],
"@types/pg": ["@types/pg@8.20.0", "", { "dependencies": { "@types/node": "*", "pg-protocol": "*", "pg-types": "^2.2.0" } }, "sha512-bEPFOaMAHTEP1EzpvHTbmwR8UsFyHSKsRisLIHVMXnpNefSbGA1bD6CVy+qKjGSqmZqNqBDV2azOBo8TgkcVow=="],
"@types/react": ["@types/react@19.2.16", "", { "dependencies": { "csstype": "^3.2.2" } }, "sha512-esJiCAnl0kfpNdE69f3So4WJUXy95dLZydX0KwK46riIHDzHM7O9Vtf9xCHW0PXIqvgqNrswl522kA/5yx+F4w=="],
@@ -1425,6 +1479,10 @@
"brace-expansion": ["brace-expansion@5.0.6", "", { "dependencies": { "balanced-match": "^4.0.2" } }, "sha512-kLpxurY4Z4r9sgMsyG0Z9uzsBlgiU/EFKhj/h91/8yHu0edo7XuixOIH3VcJ8kkxs6/jPzoI6U9Vj3WqbMQ94g=="],
"brotli": ["brotli@1.3.3", "", { "dependencies": { "base64-js": "^1.1.2" } }, "sha512-oTKjJdShmDuGW94SyyaoQvAjf30dZaHnjJ8uAF+u2/vGJkJbJPJAT1gDiOJP5v1Zb6f9KEyW/1HpuaWIXtGHPg=="],
"browserify-zlib": ["browserify-zlib@0.2.0", "", { "dependencies": { "pako": "~1.0.5" } }, "sha512-Z942RysHXmJrhqk88FmKBVq/v5tqmSkDz7p54G/MGyjMnCFFnC79XWNbg+Vta8W6Wb2qtSZTSxIGkJrRpCFEiA=="],
"browserslist": ["browserslist@4.28.2", "", { "dependencies": { "baseline-browser-mapping": "^2.10.12", "caniuse-lite": "^1.0.30001782", "electron-to-chromium": "^1.5.328", "node-releases": "^2.0.36", "update-browserslist-db": "^1.2.3" }, "bin": { "browserslist": "cli.js" } }, "sha512-48xSriZYYg+8qXna9kwqjIVzuQxi+KYWp2+5nCYnYKPTr0LvD89Jqk2Or5ogxz0NUMfIjhh2lIUX/LyX9B4oIg=="],
"buffer": ["buffer@6.0.3", "", { "dependencies": { "base64-js": "^1.3.1", "ieee754": "^1.2.1" } }, "sha512-FTiCpNxtwiZZHEZbcbTIcZjERVICn9yq/pDFkTl95/AxzD1naBctN7YO68riM/gLSDY7sdrMby8hofADYuuqOA=="],
@@ -1483,6 +1541,8 @@
"cliui": ["cliui@8.0.1", "", { "dependencies": { "string-width": "^4.2.0", "strip-ansi": "^6.0.1", "wrap-ansi": "^7.0.0" } }, "sha512-BSeNnyus75C4//NQ9gQt1/csTXyo/8Sb+afLAkzAptFuMsod9HFokGNudZpi/oQV73hnVK+sR+5PVRMd+Dr7YQ=="],
"clone": ["clone@2.1.2", "", {}, "sha512-3Pe/CF1Nn94hyhIYpjtiLhdCoEoz0DqQ+988E9gmeEdQZlojxnOb74wctFyuwWQHzqyf9X7C7MG8juUpqBJT8w=="],
"clsx": ["clsx@2.1.1", "", {}, "sha512-eYm0QWBtUrBWZWG0d386OGAw16Z995PiOVo2B7bjWSbHedGl5e0ZWaq65kOGgUSNesEIDkB9ISbTg/JK9dhCZA=="],
"cmdk": ["cmdk@1.1.1", "", { "dependencies": { "@radix-ui/react-compose-refs": "^1.1.1", "@radix-ui/react-dialog": "^1.1.6", "@radix-ui/react-id": "^1.1.0", "@radix-ui/react-primitive": "^2.0.2" }, "peerDependencies": { "react": "^18 || ^19 || ^19.0.0-rc", "react-dom": "^18 || ^19 || ^19.0.0-rc" } }, "sha512-Vsv7kFaXm+ptHDMZ7izaRsP70GgrW9NBNGswt9OZaVBLlE0SNpDq8eu/VGXyF9r7M0azK3Wy7OlYXsuyYLFzHg=="],
@@ -1523,6 +1583,8 @@
"csstype": ["csstype@3.2.3", "", {}, "sha512-z1HGKcYy2xA8AGQfwrn0PAy+PB7X/GSj3UVJW9qKyn43xWa+gl5nXmU4qqLMRzWVLFC8KusUX8T/0kCiOYpAIQ=="],
"csv-parse": ["csv-parse@7.0.1", "", {}, "sha512-+2z7Ar0APQ7Uu6fX4cn+pitRmxjZ1WPBcGmZFKmA74FCyi7Et/XZx8cjNQ5CjbZ4HCOxXCOpRBYvYH08Qa003A=="],
"d3-array": ["d3-array@3.2.4", "", { "dependencies": { "internmap": "1 - 2" } }, "sha512-tdQAmyA18i4J7wprpYq8ClcxZy3SC31QMeByyCFyRt7BVHdREQZ5lpzoe5mFEYZUWe+oq8HBvk9JjpibyEV4Jg=="],
"d3-color": ["d3-color@3.1.0", "", {}, "sha512-zg/chbXyeBtMQ1LbD/WSoW2DpC3I0mpmPdW+ynRTj/x2DAWYrIY7qeZIHidozwV24m4iavr15lNwIwLxRmOxhA=="],
@@ -1587,6 +1649,8 @@
"devlop": ["devlop@1.1.0", "", { "dependencies": { "dequal": "^2.0.0" } }, "sha512-RWmIqhcFf1lRYBvNmr7qTNuyCt/7/ns2jbpp1+PalgE/rDQcBT0fioSMUpJ93irlUhC5hrg4cYqe6U+0ImW0rA=="],
"dfa": ["dfa@1.2.0", "", {}, "sha512-ED3jP8saaweFTjeGX8HQPjeC1YYyZs98jGNZx6IiBvxW7JG5v492kamAQB3m2wop07CvU/RQmzcKr6bgcC5D/Q=="],
"diff": ["diff@8.0.4", "", {}, "sha512-DPi0FmjiSU5EvQV0++GFDOJ9ASQUVFh5kD+OzOnYdi7n3Wpm9hWWGfB/O2blfHcMVTL5WkQXSnRiK9makhrcnw=="],
"docker-compose": ["docker-compose@1.4.2", "", { "dependencies": { "yaml": "^2.2.2" } }, "sha512-rPHigTKGaEHpkUmfd69QgaOp+Os5vGJwG/Ry8lcr8W/382AmI+z/D7qoa9BybKIkqNppaIbs8RYeHSevdQjWww=="],
@@ -1739,6 +1803,8 @@
"flatted": ["flatted@3.4.2", "", {}, "sha512-PjDse7RzhcPkIJwy5t7KPWQSZ9cAbzQXcafsetQoD7sOJRQlGikNbx7yZp2OotDnJyrDcbyRq3Ttb18iYOqkxA=="],
"fontkit": ["fontkit@2.0.4", "", { "dependencies": { "@swc/helpers": "^0.5.12", "brotli": "^1.3.2", "clone": "^2.1.2", "dfa": "^1.2.0", "fast-deep-equal": "^3.1.3", "restructure": "^3.0.0", "tiny-inflate": "^1.0.3", "unicode-properties": "^1.4.0", "unicode-trie": "^2.0.0" } }, "sha512-syetQadaUEDNdxdugga9CpEYVaQIxOwk7GlwZWWZ19//qW4zE5bknOKeMBDYAASwnpaSHKJITRLMF9m1fp3s6g=="],
"for-each": ["for-each@0.3.5", "", { "dependencies": { "is-callable": "^1.2.7" } }, "sha512-dKx12eRCVIzqCxFGplyFKJMPvLEWgmNtUrpTiJIR5u97zEhRG8ySrtboPHZXx7daLxQVrl643cTzbab2tkQjxg=="],
"foreground-child": ["foreground-child@3.3.1", "", { "dependencies": { "cross-spawn": "^7.0.6", "signal-exit": "^4.0.1" } }, "sha512-gIXjKqtFuWEgzFRJA9WCQeSJLZDjgJUOMCMzxtvFq/37KojM1BFGufqsCy0r4qSQmYLsZYMeyRqzIWOMup03sw=="],
@@ -1965,6 +2031,8 @@
"joycon": ["joycon@3.1.1", "", {}, "sha512-34wB/Y7MW7bzjKRjUKTa46I2Z7eV62Rkhva+KkopW7Qvv/OSWBqvkSY7vusOPrNuZcUG3tApvdVgNB8POj3SPw=="],
"js-md5": ["js-md5@0.8.3", "", {}, "sha512-qR0HB5uP6wCuRMrWPTrkMaev7MJZwJuuw4fnwAzRgP4J4/F8RwtodOKpGp4XpqsLBFzzgqIO42efFAyz2Et6KQ=="],
"js-sha256": ["js-sha256@0.11.1", "", {}, "sha512-o6WSo/LUvY2uC4j7mO50a2ms7E/EAdbP0swigLV+nzHKTTaYnaLIWJ02VdXrsJX0vGedDESQnLsOekr94ryfjg=="],
"js-tokens": ["js-tokens@4.0.0", "", {}, "sha512-RdJUflcE3cUzKiMqQgsCu06FPu9UdIJO0beYbPhHN4k6apgJtifcoCtT9bcxOpYBtpD2kCM6Sbzg4CausW/PKQ=="],
@@ -2027,6 +2095,8 @@
"lightningcss-win32-x64-msvc": ["lightningcss-win32-x64-msvc@1.32.0", "", { "os": "win32", "cpu": "x64" }, "sha512-Amq9B/SoZYdDi1kFrojnoqPLxYhQ4Wo5XiL8EVJrVsB8ARoC1PWW6VGtT0WKCemjy8aC+louJnjS7U18x3b06Q=="],
"linebreak": ["linebreak@1.1.0", "", { "dependencies": { "base64-js": "0.0.8", "unicode-trie": "^2.0.0" } }, "sha512-MHp03UImeVhB7XZtjd0E4n6+3xr5Dq/9xI/5FptGk5FrbDR3zagPa2DS6U8ks/3HjbKWG9Q1M2ufOzxV2qLYSQ=="],
"loader-runner": ["loader-runner@4.3.2", "", {}, "sha512-DFEqQ3ihfS9blba08cLfYf1NRAIEm+dDjic073DRDc3/JspI/8wYmtDsHwd3+4hwvdxSK7PGaElfTmm0awWJ4w=="],
"locate-path": ["locate-path@6.0.0", "", { "dependencies": { "p-locate": "^5.0.0" } }, "sha512-iPZK6eYjbxRu3uB4/WZ3EsEIMJFMqAoopl3R+zuq0UjcAm/MO6KCweDgPfP3elTztoKP3KtnVHxTn2NHBSDVUw=="],
@@ -2203,6 +2273,8 @@
"package-json-from-dist": ["package-json-from-dist@1.0.1", "", {}, "sha512-UEZIS3/by4OC8vL3P2dTXRETpebLI2NiI5vIrjaD/5UtrkFX/tNbwjTSRAGC/+7CAo2pIcBaRgWmcBBHcsaCIw=="],
"pako": ["pako@0.2.9", "", {}, "sha512-NUcwaKxUxWrZLpDG+z/xZaCgQITkA/Dv4V/T6bw7VON6l1Xz/VnrBqrYjZQ12TamKHzITTfOEIYUj48y2KXImA=="],
"parse-entities": ["parse-entities@4.0.2", "", { "dependencies": { "@types/unist": "^2.0.0", "character-entities-legacy": "^3.0.0", "character-reference-invalid": "^2.0.0", "decode-named-character-reference": "^1.0.0", "is-alphanumerical": "^2.0.0", "is-decimal": "^2.0.0", "is-hexadecimal": "^2.0.0" } }, "sha512-GG2AQYWoLgL877gQIKeRPGO1xF9+eG1ujIb5soS5gPvLQ1y2o8FL90w2QWNdf9I361Mpp7726c+lj3U0qK1uGw=="],
"partial-json": ["partial-json@0.1.7", "", {}, "sha512-Njv/59hHaokb/hRUjce3Hdv12wd60MtM9Z5Olmn+nehe0QDAsRtRbJPvJ0Z91TusF0SuZRIvnM+S4l6EIP8leA=="],
@@ -2219,6 +2291,10 @@
"pathe": ["pathe@2.0.3", "", {}, "sha512-WUjGcAqP1gQacoQe+OBJsFA7Ld4DyXuUIjZ5cc75cLHvJ7dtNsTugphxIADwspS+AraAUePCKrSVtPLFj/F88w=="],
"pdfjs-dist": ["pdfjs-dist@6.1.200", "", { "optionalDependencies": { "@napi-rs/canvas": "^1.0.0" } }, "sha512-o8MolyzirkkLrcdsae/HEOiIcXWI7DS5zGpvqW8xTC2YUsW30rltFw2bDGvw/fskUdEMrQm2br68jzDS5BH2vw=="],
"pdfkit": ["pdfkit@0.19.1", "", { "dependencies": { "@noble/ciphers": "^1.0.0", "@noble/hashes": "^1.6.0", "fontkit": "^2.0.4", "js-md5": "^0.8.3", "linebreak": "^1.1.0", "png-js": "^1.1.0" } }, "sha512-6Gzk+wDwTs4VSxsR5rCMTnIl5nlmkye1oWB0l2hDB1EX6ZNSIBroKQEv+2+fPPn+stVjyqzmsqRJVDfB9fo5DA=="],
"pend": ["pend@1.2.0", "", {}, "sha512-F3asv42UuXchdzt+xXqfW1OGlVBe+mxa2mqI0pg5yAHZPvFmY3Y6drSf/GQ1A86WgWEN9Kzh/WrgKa6iGcHXLg=="],
"perfect-debounce": ["perfect-debounce@2.1.0", "", {}, "sha512-LjgdTytVFXeUgtHZr9WYViYSM/g8MkcTPYDlPa3cDqMirHjKiSZPYd6DoL7pK8AJQr+uWkQvCjHNdiMqsrJs+g=="],
@@ -2259,6 +2335,8 @@
"playwright-core": ["playwright-core@1.61.1", "", { "bin": { "playwright-core": "cli.js" } }, "sha512-h7Qlt6m4REp25qvIdvbDtVmD4LqVXfpRxhORv9L0jzETM05p4fuPJ3dKyuSXQxDSbXnmS79HAgi9589lGSpLkg=="],
"png-js": ["png-js@1.1.0", "", { "dependencies": { "browserify-zlib": "^0.2.0" } }, "sha512-PM/uYGzGdNSzqeOgly68+6wKQDL1SY0a/N+OEa/+br6LnHWOAJB0Npiamnodfq3jd2LS/i2fMeOKSAILjA+m5Q=="],
"possible-typed-array-names": ["possible-typed-array-names@1.1.0", "", {}, "sha512-/+5VFTchJDoVj3bhoqi6UeymcD00DAwb1nJwamzPvHEszJ4FpF6SNNbUbOS8yI56qHzdV8eK0qEfOSiodkTdxg=="],
"postcss": ["postcss@8.5.15", "", { "dependencies": { "nanoid": "^3.3.12", "picocolors": "^1.1.1", "source-map-js": "^1.2.1" } }, "sha512-FfR8sjd4em2T6fb3I2MwAJU7HWVMr9zba+enmQeeWFfCbm+UOC/0X4DS8XtpUTMwWMGbjKYP7xjfNekzyGmB3A=="],
@@ -2381,6 +2459,8 @@
"resolve": ["resolve@2.0.0-next.7", "", { "dependencies": { "es-errors": "^1.3.0", "is-core-module": "^2.16.2", "node-exports-info": "^1.6.0", "object-keys": "^1.1.1", "path-parse": "^1.0.7", "supports-preserve-symlinks-flag": "^1.0.0" }, "bin": { "resolve": "bin/resolve" } }, "sha512-tqt+NBWwyaMgw3zDsnygx4CByWjQEJHOPMdslYhppaQSJUtL/D4JO9CcBBlhPoI8lz9oJIDXkwXfhF4aWqP8xQ=="],
"restructure": ["restructure@3.0.2", "", {}, "sha512-gSfoiOEA0VPE6Tukkrr7I0RBdE0s7H1eFCDBk05l1KIQT1UIKNc5JZy6jdyW6eYH3aR3g5b3PuL77rq0hvwtAw=="],
"retry": ["retry@0.12.0", "", {}, "sha512-9LkiTwjUh6rT555DtE9rTX+BKByPfrMzEAtnlEtdEwr3Nkffwiihqe2bWADg+OQRjt9gl6ICdmB/ZFDCGAtSow=="],
"rollup": ["rollup@4.61.0", "", { "dependencies": { "@types/estree": "1.0.9" }, "optionalDependencies": { "@rollup/rollup-android-arm-eabi": "4.61.0", "@rollup/rollup-android-arm64": "4.61.0", "@rollup/rollup-darwin-arm64": "4.61.0", "@rollup/rollup-darwin-x64": "4.61.0", "@rollup/rollup-freebsd-arm64": "4.61.0", "@rollup/rollup-freebsd-x64": "4.61.0", "@rollup/rollup-linux-arm-gnueabihf": "4.61.0", "@rollup/rollup-linux-arm-musleabihf": "4.61.0", "@rollup/rollup-linux-arm64-gnu": "4.61.0", "@rollup/rollup-linux-arm64-musl": "4.61.0", "@rollup/rollup-linux-loong64-gnu": "4.61.0", "@rollup/rollup-linux-loong64-musl": "4.61.0", "@rollup/rollup-linux-ppc64-gnu": "4.61.0", "@rollup/rollup-linux-ppc64-musl": "4.61.0", "@rollup/rollup-linux-riscv64-gnu": "4.61.0", "@rollup/rollup-linux-riscv64-musl": "4.61.0", "@rollup/rollup-linux-s390x-gnu": "4.61.0", "@rollup/rollup-linux-x64-gnu": "4.61.0", "@rollup/rollup-linux-x64-musl": "4.61.0", "@rollup/rollup-openbsd-x64": "4.61.0", "@rollup/rollup-openharmony-arm64": "4.61.0", "@rollup/rollup-win32-arm64-msvc": "4.61.0", "@rollup/rollup-win32-ia32-msvc": "4.61.0", "@rollup/rollup-win32-x64-gnu": "4.61.0", "@rollup/rollup-win32-x64-msvc": "4.61.0", "fsevents": "~2.3.2" }, "bin": { "rollup": "dist/bin/rollup" } }, "sha512-T9mWdbWfQtp0B5lv/HX+wrhYsmXRlcWnXXmJbXqKJhlRaoS6KMhq0gpyzW4UJfclcxrEdLnTgjT2NjruLONu0g=="],
@@ -2543,6 +2623,8 @@
"thread-stream": ["thread-stream@4.2.0", "", { "dependencies": { "real-require": "^1.0.0" } }, "sha512-e2zZ96wSChazBsbENf/Pcm/4swHt2cEKQ92rhUjkL9GCKiTDJIaTBenjE/m9DXi0QBmTMDkFDdOomUy20A1tDQ=="],
"tiny-inflate": ["tiny-inflate@1.0.3", "", {}, "sha512-pkY1fj1cKHb2seWDy0B16HeWyczlJA9/WW3u3c4z/NiWDsO3DOU5D7nhTLE9CF0yXv/QZFY7sEJmj24dK+Rrqw=="],
"tiny-invariant": ["tiny-invariant@1.3.3", "", {}, "sha512-+FbBPE1o9QAYvviau/qC5SE3caw21q3xkvWKBtja5vgqOWIHHJ3ioaq1VPfn/Szqctz2bU/oYeKd9/z5BL+PVg=="],
"tinybench": ["tinybench@2.9.0", "", {}, "sha512-0+DUvqWMValLmha6lr4kD8iAMK1HzV0/aKnCtWb9v9641TnP/MFb7Pc2bxoxQjTXAErryXVgUOfv2YqNllqGeg=="],
@@ -2599,6 +2681,10 @@
"undici-types": ["undici-types@7.24.6", "", {}, "sha512-WRNW+sJgj5OBN4/0JpHFqtqzhpbnV0GuB+OozA9gCL7a993SmU+1JBZCzLNxYsbMfIeDL+lTsphD5jN5N+n0zg=="],
"unicode-properties": ["unicode-properties@1.4.1", "", { "dependencies": { "base64-js": "^1.3.0", "unicode-trie": "^2.0.0" } }, "sha512-CLjCCLQ6UuMxWnbIylkisbRj31qxHPAurvena/0iwSVbQ2G1VY5/HjV0IRabOEbDHlzZlRdCrD4NhB0JtU40Pg=="],
"unicode-trie": ["unicode-trie@2.0.0", "", { "dependencies": { "pako": "^0.2.5", "tiny-inflate": "^1.0.0" } }, "sha512-x7bc76x0bm4prf1VLg79uhAzKw8DVboClSN5VxJuQ+LKDOVEW9CdH+VY7SP+vX7xCYQqzzgQpFqz15zeLvAtZQ=="],
"unified": ["unified@11.0.5", "", { "dependencies": { "@types/unist": "^3.0.0", "bail": "^2.0.0", "devlop": "^1.0.0", "extend": "^3.0.0", "is-plain-obj": "^4.0.0", "trough": "^2.0.0", "vfile": "^6.0.0" } }, "sha512-xKvGhPWw3k84Qjh8bI3ZeJjqnyadK+GEFtazSfZv/rKeTkTjOJho6mFqh2SM96iIcZokxiOpg78GazTSg8+KHA=="],
"unionfs": ["unionfs@4.6.0", "", { "dependencies": { "fs-monkey": "^1.0.0" } }, "sha512-fJAy3gTHjFi5S3TP5EGdjs/OUMFFvI/ady3T8qVuZfkv8Qi8prV/Q8BuFEgODJslhZTT2z2qdD2lGdee9qjEnA=="],
@@ -3003,6 +3089,8 @@
"bl/readable-stream": ["readable-stream@3.6.2", "", { "dependencies": { "inherits": "^2.0.3", "string_decoder": "^1.1.1", "util-deprecate": "^1.0.1" } }, "sha512-9u/sniCrY3D5WdsERHzHE4G2YCXqoG5FTHUiCC4SIbr6XcLZBY05ya9EKjYek9O5xOAwjGq+1JdGBAS7Q9ScoA=="],
"browserify-zlib/pako": ["pako@1.0.11", "", {}, "sha512-4hLB8Py4zZce5s4yd9XzopqwVv/yGNhV1Bl8NTmCq1763HeK2+EwVTv+leGeL13Dnh2wfbqowVPXCIO0z4taYw=="],
"bun-types/@types/node": ["@types/node@22.19.19", "", { "dependencies": { "undici-types": "~6.21.0" } }, "sha512-dyh/xO2Fh5bYrfWaaqGrRQQGkNdmYw6AmaAUvYeUMNTWQtvb796ikLdmTchRmOlOiIJ1TDXfWgVx1QkUlQ6Hew=="],
"c12/dotenv": ["dotenv@17.4.2", "", {}, "sha512-nI4U3TottKAcAD9LLud4Cb7b2QztQMUEfHbvhTH09bqXTxnSie8WnjPALV/WMCrJZ6UV/qHJ6L03OqO3LcdYZw=="],
@@ -3031,12 +3119,18 @@
"lazystream/readable-stream": ["readable-stream@2.3.8", "", { "dependencies": { "core-util-is": "~1.0.0", "inherits": "~2.0.3", "isarray": "~1.0.0", "process-nextick-args": "~2.0.0", "safe-buffer": "~5.1.1", "string_decoder": "~1.1.1", "util-deprecate": "~1.0.1" } }, "sha512-8p0AUk4XODgIewSi0l8Epjs+EVnWiK7NoDIEGU0HhE7+ZyY8D1IMY7odu5lRrFXGg71L15KG8QrPmum45RTtdA=="],
"linebreak/base64-js": ["base64-js@0.0.8", "", {}, "sha512-3XSA2cR/h/73EzlXXdU6YNycmYI7+kicTxks4eJg2g39biHR84slg2+des+p7iHYhbRg/udIS4TD53WabcOUkw=="],
"p-retry/retry": ["retry@0.13.1", "", {}, "sha512-XQBQ3I8W1Cge0Seh+6gjj03LbmRFWuoszgK9ooCpwYIrhhoO80pfq4cUkU5DkknwfOfFteRwlZ56PYOGYyFWdg=="],
"parse-entities/@types/unist": ["@types/unist@2.0.11", "", {}, "sha512-CmBKiL6NNo/OqgmMn95Fk9Whlp2mtvIv+KNpQKN2F4SjvrEesubTRWGYSg+BnWZOnlCaSTU1sMpsBOzgbYhnsA=="],
"path-scurry/lru-cache": ["lru-cache@10.4.3", "", {}, "sha512-JNAzZcXrCt42VGLuYz0zfAzDfAvJWW6AfYlDBQyDV5DClI2m5sAmK+OIO7s59XfsRsWHp02jAJrRadPRGTt6SQ=="],
"pdfkit/@noble/ciphers": ["@noble/ciphers@1.3.0", "", {}, "sha512-2I0gnIVPtfnMw9ee9h1dJG7tp81+8Ob3OJb3Mv37rx5L40/b0i7djjCVvGOVqc9AEIQyvyu1i6ypKdFw8R8gQw=="],
"pdfkit/@noble/hashes": ["@noble/hashes@1.8.0", "", {}, "sha512-jCs9ldd7NwzpgXDIf6P3+NrHh9/sD6CQdxHyjQI+h/6rDNo88ypBxxz45UDuZHz9r3tNz7N/VInSVoVdtXEI4A=="],
"pg-types/postgres-array": ["postgres-array@2.0.0", "", {}, "sha512-VpZrUqU5A69eQyW2c5CA1jtLecCsN2U/bD6VilrFDWq5+5UIEVO7nazS3TEcHf1zuPYO/sqGvUvW62g86RXZuA=="],
"pretty-format/react-is": ["react-is@17.0.2", "", {}, "sha512-w2GsyukL62IJnlaff/nRegPQR94C/XXamvMWmSHRJ4y7Ts/4ocGRmTHvOs8PSE6pB3dWOrD/nueuU5sduBsQ4w=="],
@@ -3153,6 +3247,8 @@
"yauzl/buffer-crc32": ["buffer-crc32@0.2.13", "", {}, "sha512-VO9Ht/+p3SN7SKWqcrgEzjGbRSJYTx+Q1pTQC0wrWqHx0vpJraQ6GtHx8tvcg1rlK1byhU5gccxgOgj7B0TDkQ=="],
"@earendil-works/pi-agent-core/@earendil-works/pi-ai/@mistralai/mistralai": ["@mistralai/mistralai@2.2.1", "", { "dependencies": { "ws": "^8.18.0", "zod": "^3.25.0 || ^4.0.0", "zod-to-json-schema": "^3.25.0" } }, "sha512-uKU8CZmL2RzYKmplsU01hii4p3pe4HqJefpWNRWXm1Tcm0Sm4xXfwSLIy4k7ZCPlbETCGcp69E7hZs+WOJ5itQ=="],
"@isaacs/cliui/string-width/emoji-regex": ["emoji-regex@9.2.2", "", {}, "sha512-L18DaJsXSUk2+42pv8mLs5jJT2hqFkFE4j21wOmgbUqsZ2hL72NsUU785g9RXgo3s0ZNgVl42TiHp3ZtOv/Vyg=="],
"@isaacs/cliui/strip-ansi/ansi-regex": ["ansi-regex@6.2.2", "", {}, "sha512-Bq3SmSpyFHaWjPk8If9yc6svM8c56dB5BAtW4Qbw5jHTwwXXcTLoRMkpDJp6VL0XzlWaCHTXrkFURMYmD0sLqg=="],
-1
View File
@@ -45,7 +45,6 @@ Shumai supports storing files locally or on S3-compatible endpoints.
| `S3_ACCESS_KEY_ID` | Access key credentials for S3. | `minioadmin` (default) |
| `S3_SECRET_ACCESS_KEY` | Secret key credentials for S3. | `minioadmin` (default) |
| `S3_REGION` | S3 storage region. | `auto` (default) |
| `S3_USE_SSL` | Set to `true` to use HTTPS for S3 endpoints. | `false` (default) |
| `PRESIGNED_URL_EXPIRES_IN` | Number of hours for which generated presigned download/upload links are valid. | `5` (default) |
### Local Storage Port Configuration
+6 -1
View File
@@ -16,7 +16,11 @@
"groups": [
{
"group": "Getting Started",
"pages": ["getting-started/overview", "getting-started/quickstart"]
"pages": [
"getting-started/overview",
"getting-started/quickstart",
"getting-started/s3-storage"
]
},
{
"group": "Using Shumai",
@@ -39,6 +43,7 @@
"using-shumai/agent-providers",
"using-shumai/agents",
"using-shumai/agent-skills",
"using-shumai/agent-tools",
"using-shumai/agent-sandbox"
]
},
+125
View File
@@ -0,0 +1,125 @@
---
title: S3 Compatible Storage
description: Configure AWS S3 or any S3-compatible storage backend (Cloudflare R2, MinIO, Google Cloud Storage, etc.) for Shumai.
---
Shumai uses Bun's high-performance native S3 client to store and serve creative assets. By utilizing standard S3 APIs, Shumai can connect to any S3-compatible storage provider, including AWS S3, Cloudflare R2, MinIO, Google Cloud Storage, and more.
Using external S3 storage is highly recommended for production teams. It allows you to separate storage from the application server, scale storage capacity independently, and support distributed media transcoding.
---
## Configuration Variables
To enable S3-compatible storage, update the following environment variables in your `.env` file or Docker Compose configuration:
| Variable | Description | Required | Default / Example |
| :------------------------- | :------------------------------------------------------------------------ | :------- | :----------------------------------- |
| `STORAGE_BACKEND` | Set to `s3` to enable S3 compatible storage. | Yes | `s3` |
| `S3_BUCKET` | The name of the S3 bucket where files will be stored. | Yes | `shumai-bucket` |
| `S3_REGION` | The region of your S3 bucket. | No | `auto` |
| `S3_ACCESS_KEY_ID` | The access key ID for your storage service. | Yes | `your-access-key-id` |
| `S3_SECRET_ACCESS_KEY` | The secret access key for your storage service. | Yes | `your-secret-access-key` |
| `AWS_ENDPOINT_URL_S3` | The endpoint URL of your S3 compatible storage provider. | Yes | `https://s3.us-east-1.amazonaws.com` |
| `PRESIGNED_URL_EXPIRES_IN` | Duration in hours before generated presigned download/upload URLs expire. | No | `5` |
---
## Provider Examples
Below are configuration templates for popular S3 providers. Add these variables to your `.env` file or under the `environment` section in your `docker-compose.yaml`.
### AWS S3
For standard AWS S3, configure the corresponding endpoint URL:
```env
STORAGE_BACKEND=s3
S3_BUCKET=your-shumai-bucket
S3_ACCESS_KEY_ID=AKIAIOSFODNN7EXAMPLE
S3_SECRET_ACCESS_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY
AWS_ENDPOINT_URL_S3=https://s3.us-east-1.amazonaws.com
```
### Cloudflare R2
Cloudflare R2 requires your Cloudflare account ID to construct the endpoint URL. Set the region to `auto`:
```env
STORAGE_BACKEND=s3
S3_BUCKET=your-shumai-bucket
S3_ACCESS_KEY_ID=your-r2-access-key-id
S3_SECRET_ACCESS_KEY=your-r2-secret-access-key
AWS_ENDPOINT_URL_S3=https://<account-id>.r2.cloudflarestorage.com
```
### MinIO (Self-Hosted)
For local development or self-hosted environments using MinIO, set the endpoint to your MinIO server URL and port (usually `9000`):
```env
STORAGE_BACKEND=s3
S3_BUCKET=shumai
S3_ACCESS_KEY_ID=minioadmin
S3_SECRET_ACCESS_KEY=minioadmin
AWS_ENDPOINT_URL_S3=http://localhost:9000
```
### Google Cloud Storage (GCS)
Google Cloud Storage provides an S3-compatible XML API. Configure your HMAC credentials and set the endpoint to standard GCS storage:
```env
STORAGE_BACKEND=s3
S3_BUCKET=your-gcs-bucket
S3_ACCESS_KEY_ID=GOOG1EXAMPLETOCKENKEY
S3_SECRET_ACCESS_KEY=your-gcs-hmac-secret-access-key
AWS_ENDPOINT_URL_S3=https://storage.googleapis.com
```
### DigitalOcean Spaces
For DigitalOcean Spaces, set the endpoint URL to match your space's region:
```env
STORAGE_BACKEND=s3
S3_BUCKET=your-spaces-bucket
S3_ACCESS_KEY_ID=your-spaces-access-key-id
S3_SECRET_ACCESS_KEY=your-spaces-secret-access-key
AWS_ENDPOINT_URL_S3=https://nyc3.digitaloceanspaces.com
```
### Supabase Storage
To connect to Supabase Storage via S3, ensure you have enabled S3 connection support in the Supabase Dashboard (`Settings > Storage > Enable connection via S3 protocol`). Copy the endpoint URL and region details from the same settings page:
```env
STORAGE_BACKEND=s3
S3_BUCKET=your-supabase-bucket
S3_REGION=us-east-1
S3_ACCESS_KEY_ID=your-supabase-access-key-id
S3_SECRET_ACCESS_KEY=your-supabase-secret-access-key
AWS_ENDPOINT_URL_S3=https://<account-id>.supabase.co/storage/v1/s3
```
---
## CORS Configuration
Since Shumai uploads files directly from the user's browser using presigned URLs, you must configure **Cross-Origin Resource Sharing (CORS)** on your storage bucket. Without CORS configuration, uploads will fail with network or origin errors.
Ensure your bucket CORS policy allows the following rules from your Shumai application URL (e.g., `http://localhost:3000` or `https://app.shumai.one`):
### CORS JSON Policy Example (AWS S3, R2, GCS)
```json
[
{
"AllowedHeaders": ["*"],
"AllowedMethods": ["GET", "PUT", "POST", "DELETE", "HEAD"],
"AllowedOrigins": ["http://localhost:3000", "https://your-shumai-domain.com"],
"ExposeHeaders": ["ETag"],
"MaxAgeSeconds": 3000
}
]
```
+73
View File
@@ -0,0 +1,73 @@
---
title: Agent Tools
description: Configure the tools and capabilities permitted for each agent.
---
Shumai agents are equipped with built-in tools that allow them to interact with the file system, view files, extract video frames, and execute commands.
By default, **all tools are enabled** for new agents. Team Owners can deny specific tools to restrict agent capabilities.
<Warning>
**Caution:** Disabling tools can cause agents or skills to fail. For example, disabling the
**Sandboxed Terminal** tool will break almost all custom skills that rely on executing commands or
script environments. Disable tools only when you have a specific security policy or isolation
requirement.
</Warning>
---
## Available Tools
The following tools can be individually toggled in the agent configuration dialog:
### 1. Sandboxed Terminal (`bash`)
- **Description:** Allows the agent to run terminal/shell commands in a highly isolated environment.
- **Why it matters:** Essential for running custom scripts, processing files on disk, or using standard Unix utilities. If disabled, the agent won't be able to run command-line code.
### 2. Read Skills (`read_skill`)
- **Description:** Allows the agent to load and read skill definitions to perform specialized actions.
- **Why it matters:** Enables the agent to inspect the instructions/configurations of custom skills registered in your workspace before executing them.
### 3. Analyze Image (`analyze_image`)
- **Description:** Allows the agent to view and analyze the visual contents of an image file.
- **Why it matters:** Enables multimodal agents to "see" image assets uploaded to Shumai, allowing them to answer comments or write metadata about images.
### 4. Video Frame Extraction (`screenshot`)
- **Description:** Allows the agent to extract still image frames from video files for analysis.
- **Why it matters:** Enables the agent to perform frame-by-frame analysis on videos (e.g. for comments, annotation, or search).
### 5. Browse Shumai (`list_assets`)
- **Description:** Allows the agent to list folders and browse files in Shumai.
- **Why it matters:** Gives the agent visual mapping of the project structure so it can search for other files or locate related assets.
### 6. Create Folder (`create_folder`)
- **Description:** Allows the agent to create new folders in Shumai.
- **Why it matters:** Needed if the agent organizes outputs or places files in custom subdirectory folders.
### 7. Create File (`create_file`)
- **Description:** Allows the agent to create and upload new files to Shumai.
- **Why it matters:** Allows the agent to generate new assets (e.g. text logs, transcoded clips, reports) and save them directly back into your project.
### 8. Create Version (`create_version`)
- **Description:** Allows the agent to upload and stack a new version of an existing file in Shumai.
- **Why it matters:** Allows the agent to update files in a clean, non-destructive way by stacking revisions rather than creating duplicate files.
---
## Restricting Tools
To change which tools an agent can use:
1. Navigate to **Team Settings** > **Agents** (Note: Only Team Owners can view and modify these settings).
2. Click on the agent you want to edit.
3. Scroll down below the **Skills** section to find the **Tools** configuration panel.
4. Toggle the switches next to the tools to enable or disable them.
5. Click **Save Changes**.
+1
View File
@@ -76,6 +76,7 @@ export default tseslint.config(
'**/data/**',
'**/packages/webui/components/ui/**',
'**/packages/webui/paraglide/**',
'**/packages/webui/public/**',
],
},
js.configs.recommended,
+2 -5
View File
@@ -1,6 +1,6 @@
{
"name": "shumai",
"version": "0.1.6",
"version": "0.2.2",
"private": true,
"type": "module",
"repository": {
@@ -19,6 +19,7 @@
"lint": "eslint . --ext ts,tsx --report-unused-disable-directives --max-warnings 0",
"test": "bun --bun vitest run",
"test:e2e": "playwright test",
"test:e2e:workflow": "bun --bun vitest run -c packages/e2e/vitest.config.ts",
"test:e2e:ui": "playwright test --ui",
"test:e2e:report": "playwright show-report",
"test:e2e:install": "playwright install --with-deps",
@@ -65,9 +66,5 @@
"typescript-eslint": "8.60.1",
"vite-tsconfig-paths": "6.1.1",
"vitest": "4.1.8"
},
"dependencies": {
"@tanstack/react-virtual": "3.14.2",
"jittered-fractional-indexing": "1.0.1"
}
}
+10 -9
View File
@@ -1,25 +1,26 @@
{
"name": "@shumai/agent",
"version": "0.1.6",
"version": "0.2.2",
"private": true,
"type": "module",
"main": "./src/index.ts",
"types": "./src/index.ts",
"dependencies": {
"@shumai/db": "workspace:*",
"@shumai/dtos": "workspace:*",
"@shumai/core": "workspace:*",
"@shumai/workflow-core": "workspace:*",
"@google/genai": "1.52.0",
"@anthropic-ai/sandbox-runtime": "0.0.52",
"@earendil-works/pi-agent-core": "0.78.0",
"@earendil-works/pi-ai": "0.75.4",
"@earendil-works/pi-ai": "0.80.7",
"@google/genai": "1.52.0",
"@shumai/core": "workspace:*",
"@shumai/db": "workspace:*",
"@shumai/dtos": "workspace:*",
"@shumai/workflow-core": "workspace:*",
"@sinclair/typebox": "0.34.49",
"@temporalio/activity": "1.17.2",
"@temporalio/worker": "1.17.2",
"@temporalio/workflow": "1.17.2",
"@sinclair/typebox": "0.34.49",
"ulid": "3.0.2",
"adm-zip": "0.5.17",
"jittered-fractional-indexing": "1.0.1",
"ulid": "3.0.2",
"zod": "4.4.3"
},
"devDependencies": {
+441
View File
@@ -15,10 +15,15 @@ import {
updateAssetMetadataActivity,
getAssetPathContextActivity,
executeAgentToolActivity,
generateSessionNameActivity,
type GenerateSessionNameParams,
} from './agent'
import * as piAgent from '../index'
import { type AgentHarness, type Session } from '@earendil-works/pi-agent-core'
import { type DatabaseSessionMetadata } from '../database-session-storage'
import { triggerLocalCancel } from '@shumai/workflow-core'
import { Context } from '@temporalio/activity'
import {
prisma,
AssetType,
@@ -44,10 +49,14 @@ vi.mock('../index', async () => {
describe('Agent Activities', () => {
beforeEach(() => {
vi.clearAllMocks()
vi.spyOn(Context, 'current').mockImplementation(() => {
throw new Error('Not in Temporal')
})
})
it('should call agentChatActivity and prompt the harness', async () => {
const mockHarness = {
subscribe: vi.fn(),
prompt: vi.fn().mockResolvedValue({
content: [{ type: 'text', text: 'AI response' }],
usage: { input: 10, output: 20 },
@@ -87,8 +96,226 @@ describe('Agent Activities', () => {
expect(piAgent.createAgentSession).toHaveBeenCalled()
})
it('should subscribe to harness events and record usage for assistant message_end events', async () => {
const { aiUsageService } = await import('@shumai/core/src/ai-usage/ai-usage')
const spyRecordUsage = vi.spyOn(aiUsageService, 'recordUsage').mockResolvedValue()
let subscribeListener!: (event: unknown) => Promise<void>
const mockHarness = {
subscribe: vi.fn().mockImplementation((listener) => {
subscribeListener = listener
}),
prompt: vi.fn().mockImplementation(async () => {
// Trigger assistant message end event during turn
if (subscribeListener) {
await subscribeListener({
type: 'message_end',
message: {
role: 'assistant',
usage: {
input: 150,
output: 75,
cacheRead: 25,
totalTokens: 225,
cost: { total: 0.0015 },
},
},
})
}
return {
content: [{ type: 'text', text: 'Subscribed response' }],
usage: { input: 150, output: 75 },
}
}),
}
const mockSession = {
getEntries: vi.fn().mockResolvedValue([]),
getStorage: vi.fn().mockReturnValue({ sessionId: 'mock-session-id' }),
}
vi.mocked(piAgent.createAgentSession).mockResolvedValue({
session: mockSession as unknown as Session<DatabaseSessionMetadata>,
harness: mockHarness as unknown as AgentHarness,
})
const context = {
agent: { id: 'b1', provider: { name: 'google' }, modelRef: { modelId: 'gemini' } },
dbProviders: [],
teamSkills: [],
allowedDomains: [],
} as unknown as AgentExecutionContext
const res = await agentChatActivity({
teamId: 'team123',
agentId: 'b1',
message: 'Hi',
imageUrls: [],
projectId: 'p1',
folderId: 'f1',
sessionId: 'mock-session-id',
userId: 'user123',
context,
})
expect(res.text).toBe('Subscribed response')
expect(res.usage.inputTokens).toBe(150)
expect(res.usage.outputTokens).toBe(75)
expect(res.usage.cacheReadTokens).toBe(25)
expect(res.usage.cost).toBe(0.0015)
expect(spyRecordUsage).toHaveBeenCalledWith({
teamId: 'team123',
userId: 'user123',
inputTokens: 150,
outputTokens: 75,
cacheReadTokens: 25,
totalTokens: 225,
cost: 0.0015,
})
})
it('should register local cancellation handler and call harness.abort when cancelled in local mode', async () => {
let resolvePrompt!: (value: {
content: Array<{ type: string; text: string } | { type: string }>
usage: { input: number; output: number }
stopReason?: string
errorMessage?: string
}) => void
const promptPromise = new Promise<Parameters<typeof resolvePrompt>[0]>((resolve) => {
resolvePrompt = resolve
})
const mockHarness = {
subscribe: vi.fn(),
prompt: vi.fn().mockReturnValue(promptPromise),
abort: vi.fn(),
}
const mockSession = {
getEntries: vi.fn().mockResolvedValue([]),
getStorage: vi.fn().mockReturnValue({ sessionId: 'mock-session-id' }),
}
vi.mocked(piAgent.createAgentSession).mockResolvedValue({
session: mockSession as unknown as Session<DatabaseSessionMetadata>,
harness: mockHarness as unknown as AgentHarness,
})
const context = {
agent: { id: 'b1', provider: { name: 'google' }, modelRef: { modelId: 'gemini' } },
dbProviders: [],
teamSkills: [],
allowedDomains: [],
} as unknown as AgentExecutionContext
const executionPromise = agentChatActivity({
taskId: 'task-123',
teamId: 't1',
agentId: 'b1',
message: 'Hi',
imageUrls: [],
projectId: 'p1',
folderId: 'f1',
sessionId: 'mock-session-id',
context,
})
// Yield to let the asynchronous setup register the cancel handler
await new Promise((resolve) => setTimeout(resolve, 50))
// Simulate user cancellation
const cancelled = triggerLocalCancel('task-123')
expect(cancelled).toBe(true)
expect(mockHarness.abort).toHaveBeenCalled()
// Resolve the prompt promise to allow the activity to finish clean
resolvePrompt({
content: [{ type: 'text', text: 'AI response' }],
usage: { input: 10, output: 20 },
})
await executionPromise
})
it('should listen to Temporal cancellation signal and call harness.abort when cancelled in Temporal mode', async () => {
const mockTemporalContext = {
cancellationSignal: new EventTarget(),
heartbeat: vi.fn(),
} as unknown as Context
vi.spyOn(Context, 'current').mockReturnValue(mockTemporalContext)
let resolvePrompt!: (value: {
content: Array<{ type: string; text: string } | { type: string }>
usage: { input: number; output: number }
stopReason?: string
errorMessage?: string
}) => void
const promptPromise = new Promise<Parameters<typeof resolvePrompt>[0]>((resolve) => {
resolvePrompt = resolve
})
const mockHarness = {
subscribe: vi.fn(),
prompt: vi.fn().mockReturnValue(promptPromise),
abort: vi.fn(),
}
const mockSession = {
getEntries: vi.fn().mockResolvedValue([]),
getStorage: vi.fn().mockReturnValue({ sessionId: 'mock-session-id' }),
}
vi.mocked(piAgent.createAgentSession).mockResolvedValue({
session: mockSession as unknown as Session<DatabaseSessionMetadata>,
harness: mockHarness as unknown as AgentHarness,
})
const context = {
agent: { id: 'b1', provider: { name: 'google' }, modelRef: { modelId: 'gemini' } },
dbProviders: [],
teamSkills: [],
allowedDomains: [],
} as unknown as AgentExecutionContext
const addEventListenerSpy = vi.spyOn(mockTemporalContext.cancellationSignal, 'addEventListener')
const removeEventListenerSpy = vi.spyOn(
mockTemporalContext.cancellationSignal,
'removeEventListener',
)
const executionPromise = agentChatActivity({
teamId: 't1',
agentId: 'b1',
message: 'Hi',
imageUrls: [],
projectId: 'p1',
folderId: 'f1',
sessionId: 'mock-session-id',
context,
})
// Yield to let the asynchronous setup add the abort event listener
await new Promise((resolve) => setTimeout(resolve, 50))
expect(addEventListenerSpy).toHaveBeenCalledWith('abort', expect.any(Function), { once: true })
// Simulate Temporal cancellation event
mockTemporalContext.cancellationSignal.dispatchEvent(new Event('abort'))
expect(mockHarness.abort).toHaveBeenCalled()
// Resolve the prompt promise to allow the activity to finish clean
resolvePrompt({
content: [{ type: 'text', text: 'AI response' }],
usage: { input: 10, output: 20 },
})
await executionPromise
expect(removeEventListenerSpy).toHaveBeenCalledWith('abort', expect.any(Function))
addEventListenerSpy.mockRestore()
removeEventListenerSpy.mockRestore()
})
it('should include error message in text when stopReason is error', async () => {
const mockHarness = {
subscribe: vi.fn(),
prompt: vi.fn().mockResolvedValue({
content: [],
usage: { input: 0, output: 0 },
@@ -129,6 +356,7 @@ describe('Agent Activities', () => {
it('should pass thinkingLevel from agent config to createAgentSession', async () => {
const mockHarness = {
subscribe: vi.fn(),
prompt: vi.fn().mockResolvedValue({
content: [],
usage: { input: 0, output: 0 },
@@ -177,6 +405,7 @@ describe('Agent Activities', () => {
it('should call autofillAiActivity and run autofill tool', async () => {
const mockHarness = {
subscribe: vi.fn(),
prompt: vi.fn().mockResolvedValue({
content: [{ type: 'text', text: 'Captured' }],
usage: { input: 5, output: 5 },
@@ -451,6 +680,45 @@ describe('Agent Database Activities Integration', () => {
const fetchedComment = await getCommentActivity(comment.id)
expect(fetchedComment?.id).toBe(comment.id)
})
it('should resolve version stack asset to its latest version file', async () => {
const stack = await prisma.asset.create({
data: {
name: 'Stack',
type: AssetType.version_stack,
status: AssetStatus.uploaded,
projectId: project.id,
},
})
await prisma.asset.create({
data: {
name: 'v1.png',
type: AssetType.file,
mediaType: 'image/png',
status: AssetStatus.uploaded,
projectId: project.id,
parentId: stack.id,
sortIndex: 'a1',
},
})
const v2 = await prisma.asset.create({
data: {
name: 'v2.png',
type: AssetType.file,
mediaType: 'image/png',
status: AssetStatus.uploaded,
projectId: project.id,
parentId: stack.id,
sortIndex: 'a0',
},
})
const fetchedAsset = await getAssetActivity(stack.id)
expect(fetchedAsset?.id).toBe(v2.id)
expect(fetchedAsset?.name).toBe('v2.png')
})
})
describe('getAssetPathContextActivity', () => {
@@ -741,4 +1009,177 @@ describe('Agent Database Activities Integration', () => {
})
})
})
describe('generateSessionNameActivity', () => {
it('should generate and save session name for chat sessions', async () => {
await prisma.team.create({
data: { id: 't1', name: 'Test Team' },
})
const user = await prisma.user.create({
data: { id: 'u1', name: 'Test User', email: 'u1@test.com' },
})
const agentUser = await prisma.user.create({
data: {
id: 'agent-123',
name: 'Chat Agent User',
email: 'agent@shumai.ai',
type: 'agent',
},
})
const agent = await prisma.agent.create({
data: {
id: agentUser.id,
teamId: 't1',
type: 'chat',
config: { provider: 'openai', model: 'gpt-4' },
},
})
await prisma.agentSession.create({
data: {
id: 'session-123',
agentId: 'agent-123',
userId: user.id,
cwd: process.cwd(),
type: 'chat',
name: null,
},
})
const mockHarness = {
subscribe: vi.fn(),
prompt: vi.fn().mockResolvedValue({
content: [{ type: 'text', text: ' "Summarized Chat Title" ' }],
}),
}
vi.mocked(piAgent.createAgentSession).mockResolvedValue({
session: {} as unknown as Session<DatabaseSessionMetadata>,
harness: mockHarness as unknown as AgentHarness,
})
await generateSessionNameActivity({
teamId: 't1',
agentId: 'agent-123',
prompt: 'hello world prompt',
sessionId: 'session-123',
context: {
agent: {
...agent,
provider: { name: 'openai' },
modelRef: {
id: 'm1',
providerName: 'openai',
modelId: 'gpt-4',
name: 'GPT-4',
config: {},
createdAt: new Date(),
updatedAt: new Date(),
},
} as unknown as GenerateSessionNameParams['context']['agent'],
dbProviders: [
{
name: 'openai',
config: { api: 'openai-responses', apiKey: 'fake-key' },
models: [
{
modelId: 'gpt-4',
name: 'GPT-4',
config: {
reasoning: false,
input: ['text'],
contextWindow: 4096,
maxTokens: 4096,
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
},
},
],
},
],
},
})
const updatedSession = await prisma.agentSession.findUnique({
where: { id: 'session-123' },
})
expect(updatedSession?.name).toBe('Summarized Chat Title')
expect(mockHarness.prompt).toHaveBeenCalledWith('hello world prompt')
// Verify that transient naming session was cleaned up (deleted)
const namingSessions = await prisma.agentSession.findMany({
where: { type: 'naming' },
})
expect(namingSessions).toHaveLength(0)
})
it('should skip name generation for comment sessions or if name is already set', async () => {
await prisma.team.create({
data: { id: 't1', name: 'Test Team' },
})
const user = await prisma.user.create({
data: { id: 'u1', name: 'Test User', email: 'u1@test.com' },
})
const agentUser = await prisma.user.create({
data: {
id: 'agent-123',
name: 'Chat Agent User',
email: 'agent2@shumai.ai',
type: 'agent',
},
})
await prisma.agent.create({
data: {
id: agentUser.id,
teamId: 't1',
type: 'chat',
config: { provider: 'openai', model: 'gpt-4' },
},
})
await prisma.agentSession.create({
data: {
id: 'session-comment',
agentId: 'agent-123',
userId: user.id,
cwd: process.cwd(),
type: 'comment',
name: null,
},
})
await prisma.agentSession.create({
data: {
id: 'session-named',
agentId: 'agent-123',
userId: user.id,
cwd: process.cwd(),
type: 'chat',
name: 'Existing Name',
},
})
vi.mocked(piAgent.createAgentSession).mockClear()
await generateSessionNameActivity({
teamId: 't1',
agentId: 'agent-123',
prompt: 'hello comment',
sessionId: 'session-comment',
context: {
agent: {} as unknown as GenerateSessionNameParams['context']['agent'],
dbProviders: [],
},
})
await generateSessionNameActivity({
teamId: 't1',
agentId: 'agent-123',
prompt: 'hello named',
sessionId: 'session-named',
context: {
agent: {} as unknown as GenerateSessionNameParams['context']['agent'],
dbProviders: [],
},
})
expect(piAgent.createAgentSession).not.toHaveBeenCalled()
})
})
})
+288 -19
View File
@@ -1,29 +1,39 @@
import { createAgentSession, fieldsToTypeBoxSchema, type AutofillField } from '../index'
import { DatabaseSessionStorage } from '../database-session-storage'
import { prisma, AssetType, Prisma, type Skill } from '@shumai/db'
import { logger } from '@shumai/core/src/logger'
import { s3Service } from '@shumai/core/src/s3/s3'
import {
type AgentTool,
type AgentMessage,
type AgentTool,
type SessionTreeEntry,
} from '@earendil-works/pi-agent-core'
import { type ImageContent } from '@earendil-works/pi-ai'
import { ApplicationFailure } from '@temporalio/activity'
import { assetService } from '@shumai/core/src/asset/asset'
import { VersionStackService } from '@shumai/core/src/versionStack/versionStack'
import { uploadService } from '@shumai/core/src/upload/upload'
import { authzService, Permission, ResourceType } from '@shumai/core/src/authz/authz'
import { generateKeyBetween } from 'jittered-fractional-indexing'
import { paginateQuery, encodeCursor } from '@shumai/core/src/pagination'
import { ulid } from 'ulid'
import { logger } from '@shumai/core/src/logger'
import { metadataService } from '@shumai/core/src/metadata/metadata'
import { encodeCursor, paginateQuery } from '@shumai/core/src/pagination'
import { s3Service } from '@shumai/core/src/s3/s3'
import { uploadService } from '@shumai/core/src/upload/upload'
import { VersionStackService } from '@shumai/core/src/versionStack/versionStack'
import { AssetType, prisma, Prisma, type Skill } from '@shumai/db'
import { ApplicationFailure, Context } from '@temporalio/activity'
import { registerLocalCancelHandler, unregisterLocalCancelHandler } from '@shumai/workflow-core'
import { generateKeyBetween } from 'jittered-fractional-indexing'
import { ulid } from 'ulid'
import { DatabaseSessionStorage } from '../database-session-storage'
import {
createAgentSession,
fieldsToTypeBoxSchema,
type AutofillField,
type DbProviderInfo,
} from '../index'
import { aiUsageService } from '@shumai/core/src/ai-usage/ai-usage'
import { UpdateAssetMetadataRequest } from '@shumai/dtos'
export interface Usage {
inputTokens: number
outputTokens: number
cacheReadTokens: number
totalTokens: number
cost: number
model: string
}
@@ -48,6 +58,7 @@ export interface AgentExecutionContext {
}
async function executeAgentPrompt(params: {
taskId?: string
teamId: string
agentId: string
prompt: string
@@ -58,6 +69,7 @@ async function executeAgentPrompt(params: {
userCommentId?: string | null
tools?: AgentTool[]
context: AgentExecutionContext
attachedAssets?: Array<{ id: string; name: string; type: string }>
}): Promise<{ text: string; usage: Usage; sessionId: string }> {
const { agent, dbProviders, teamSkills, allowedDomains } = params.context
@@ -88,9 +100,8 @@ If you need to create files in the local filesystem (for example, a temporary fi
systemPrompt = `${systemPrompt}\n\nAgent Personality and Core Instructions:\n${agent.soul}`
}
if (params.agentsInstruction) {
systemPrompt = `${systemPrompt}\n\nContext and Instructions:\n${params.agentsInstruction}`
}
// Note: params.agentsInstruction is no longer appended to systemPrompt here.
// Instead, it is persisted as a custom message session entry in the conversation history below.
const modelConfig = agent.modelRef?.config
if (modelConfig?.input) {
@@ -144,9 +155,114 @@ If you need to create files in the local filesystem (for example, a temporary fi
}
}
let heartbeatInterval: ReturnType<typeof setInterval> | null = null
const onAbort = () => {
harness.abort()
}
let temporalSignal: EventTarget | undefined
try {
temporalSignal = Context.current().cancellationSignal
} catch {
// Ignore
}
if (temporalSignal) {
temporalSignal.addEventListener('abort', onAbort, { once: true })
heartbeatInterval = setInterval(() => {
try {
Context.current().heartbeat()
} catch (err) {
logger.debug({ err }, 'Failed to send activity heartbeat')
}
}, 2000)
} else if (params.taskId) {
registerLocalCancelHandler(params.taskId, () => {
harness.abort()
})
}
try {
if (params.agentsInstruction && params.agentsInstruction.trim()) {
const storage = session.getStorage()
const parentId = await storage.getLeafId()
await storage.appendEntry({
id: ulid(),
type: 'custom_message',
parentId,
timestamp: new Date().toISOString(),
customType: 'context',
content: params.agentsInstruction.trim(),
} as unknown as SessionTreeEntry)
}
if (params.attachedAssets && params.attachedAssets.length > 0) {
const storage = session.getStorage()
const parentId = await storage.getLeafId()
await storage.appendEntry({
id: ulid(),
type: 'custom',
parentId,
timestamp: new Date().toISOString(),
customType: 'context_display_info',
data: {
assets: params.attachedAssets,
},
} as unknown as SessionTreeEntry)
}
let totalInputTokens = 0
let totalOutputTokens = 0
let totalCacheReadTokens = 0
let grandTotalTokens = 0
let totalCost = 0
harness.subscribe(async (event) => {
if (event.type === 'message_end' && event.message.role === 'assistant') {
const assistantMsg = event.message
const u = assistantMsg.usage
if (u) {
const inputTokens = u.input || 0
const outputTokens = u.output || 0
const cacheReadTokens = u.cacheRead || 0
const totalTokens = u.totalTokens || inputTokens + outputTokens
const cost = u.cost?.total || 0
totalInputTokens += inputTokens
totalOutputTokens += outputTokens
totalCacheReadTokens += cacheReadTokens
grandTotalTokens += totalTokens
totalCost += cost
try {
await aiUsageService.recordUsage({
teamId: params.teamId,
userId: params.userId,
inputTokens,
outputTokens,
cacheReadTokens,
totalTokens,
cost,
})
} catch (err) {
logger.error({ err }, 'Failed to record AI usage')
}
}
}
})
const assistantMessage = await harness.prompt(params.prompt, { images: imagesToPass })
if (totalInputTokens === 0 && totalOutputTokens === 0 && assistantMessage.usage) {
totalInputTokens = assistantMessage.usage.input || 0
totalOutputTokens = assistantMessage.usage.output || 0
totalCacheReadTokens = assistantMessage.usage.cacheRead || 0
grandTotalTokens = assistantMessage.usage.totalTokens || totalInputTokens + totalOutputTokens
totalCost = assistantMessage.usage.cost?.total || 0
}
const sessionEntries = await session.getEntries()
sessionEntries.forEach((entry) => {
if (entry.type === 'message') {
@@ -179,8 +295,11 @@ If you need to create files in the local filesystem (for example, a temporary fi
const usage: Usage = {
model: modelId,
inputTokens: assistantMessage.usage?.input || 0,
outputTokens: assistantMessage.usage?.output || 0,
inputTokens: totalInputTokens,
outputTokens: totalOutputTokens,
cacheReadTokens: totalCacheReadTokens,
totalTokens: grandTotalTokens,
cost: totalCost,
}
const storage = session.getStorage()
@@ -190,11 +309,20 @@ If you need to create files in the local filesystem (for example, a temporary fi
}
return { text, usage, sessionId }
} finally {
// Cleanup handled by agent session if needed
if (heartbeatInterval) {
clearInterval(heartbeatInterval)
}
if (temporalSignal) {
temporalSignal.removeEventListener('abort', onAbort)
}
if (params.taskId) {
unregisterLocalCancelHandler(params.taskId)
}
}
}
export interface AgentChatParams {
taskId?: string
teamId: string
agentId: string
message: string
@@ -207,6 +335,7 @@ export interface AgentChatParams {
userCommentId?: string | null
explicitMention?: boolean
context: AgentExecutionContext
attachedAssets?: Array<{ id: string; name: string; type: string }>
}
export async function agentChatActivity(params: AgentChatParams) {
@@ -214,6 +343,7 @@ export async function agentChatActivity(params: AgentChatParams) {
const sessionId = params.sessionId
return executeAgentPrompt({
taskId: params.taskId,
teamId: params.teamId,
agentId: params.agentId,
prompt: cleanMessage,
@@ -223,6 +353,7 @@ export async function agentChatActivity(params: AgentChatParams) {
userId: params.userId,
userCommentId: params.userCommentId,
context: params.context,
attachedAssets: params.attachedAssets,
})
}
@@ -687,7 +818,7 @@ export async function getAgentAutofillContextActivity(params: {
}
export async function getAssetActivity(assetId: string) {
return prisma.asset.findUnique({
const asset = await prisma.asset.findUnique({
where: { id: assetId },
include: {
storageKey: true,
@@ -698,6 +829,30 @@ export async function getAssetActivity(assetId: string) {
},
},
})
if (asset && asset.type === AssetType.version_stack) {
const latestVersion = await prisma.asset.findFirst({
where: { parentId: asset.id, isDeleted: false },
orderBy: { sortIndex: 'asc' },
include: {
storageKey: true,
project: {
include: {
team: true,
},
},
},
})
if (latestVersion) {
return {
...latestVersion,
project: latestVersion.project ?? asset.project,
}
}
}
return asset
}
export async function getCommentActivity(commentId: string) {
@@ -1108,3 +1263,117 @@ export async function executeAgentToolActivity(params: ExecuteAgentToolParams):
})
}
}
export interface GenerateSessionNameParams {
teamId: string
agentId: string
prompt: string
sessionId: string
context: {
agent: Prisma.AgentGetPayload<{ include: { modelRef: true; provider: true } }>
dbProviders: DbProviderInfo[]
}
}
export async function generateSessionNameActivity(
params: GenerateSessionNameParams,
): Promise<void> {
const { sessionId, prompt, teamId, agentId } = params
const { agent, dbProviders } = params.context
// 1. Fetch current session and verify it is a chatbot session and unnamed
const sessionRecord = await prisma.agentSession.findUnique({
where: { id: sessionId },
})
if (!sessionRecord || sessionRecord.type !== 'chat' || sessionRecord.name !== null) {
logger.info({ sessionId }, 'Skipping session name generation: not an unnamed chatbot session')
return
}
if (!agent || !agent.modelRef || !agent.provider) {
logger.warn(
{ sessionId },
'Skipping session name generation: Agent, model, or provider configuration not found',
)
return
}
const providerName = agent.provider.name
const modelId = agent.modelRef.modelId
let namingSessionId: string | undefined
try {
// 2. Create transient naming session in database
const namingSession = await prisma.agentSession.create({
data: {
agentId,
userId: sessionRecord.userId,
cwd: process.cwd(),
assetId: sessionRecord.assetId,
type: 'naming',
},
})
namingSessionId = namingSession.id
const systemInstruction =
"You are a helpful assistant. Generate a short 2 to 4 words title/name for a chat session based on the user's first message. " +
'Do NOT include quotation marks, markdown formatting, or any extra conversational text. Return only the title.'
// 3. Create agent session and harness using the naming session
const { harness } = await createAgentSession({
teamId,
agentId,
providerName,
modelId,
systemPrompt: systemInstruction,
teamSkills: [],
allowedDomains: [],
sessionId: namingSessionId,
userId: sessionRecord.userId || undefined,
providers: dbProviders,
})
const result = await harness.prompt(prompt)
const resultText = result.content
.filter((c) => c.type === 'text')
.map((c) => {
if ('text' in c && typeof c.text === 'string') {
return c.text
}
return ''
})
.join('\n')
let chatName = resultText.trim()
// Clean up quotes if returned
if (
(chatName.startsWith('"') && chatName.endsWith('"')) ||
(chatName.startsWith("'") && chatName.endsWith("'"))
) {
chatName = chatName.slice(1, -1).trim()
}
if (chatName) {
await prisma.agentSession.update({
where: { id: sessionId },
data: { name: chatName },
})
logger.info({ sessionId, chatName }, 'Session name generated successfully')
}
} catch (err) {
logger.error({ sessionId, err }, 'Failed to generate session name')
} finally {
if (namingSessionId) {
await prisma.agentSession
.delete({
where: { id: namingSessionId },
})
.catch((deleteErr) => {
logger.error({ namingSessionId, deleteErr }, 'Failed to delete naming session')
})
}
}
}
+3 -2
View File
@@ -88,8 +88,9 @@ export async function generateEmbeddingActivity(params: GenerateEmbeddingParams)
outputTokens: 0,
}
const isImage = asset.mediaType.startsWith('image/')
const isVideo = asset.mediaType.startsWith('video/')
const proxyType = (asset.media as PrismaJson.MediaInfo | null)?.proxyType
const isImage = proxyType === 'image'
const isVideo = proxyType === 'video'
if (!isImage && !isVideo) {
throw ApplicationFailure.create({
@@ -2,7 +2,7 @@ import { prisma } from '@shumai/db'
import { setupTestDbHooks } from '@shumai/db/test'
import { s3Service } from '@shumai/core/src/s3/s3'
import { type SessionTreeEntry } from '@earendil-works/pi-agent-core'
import { describe, expect, it, vi } from 'vitest'
import { describe, expect, it, vi, afterEach } from 'vitest'
import { DatabaseSessionStorage } from './database-session-storage'
import { agentService } from '@shumai/core/src/agent/agent'
import { createAgentSession } from './index'
@@ -13,6 +13,10 @@ import * as screenshotModule from './tools/screenshot'
describe('DatabaseSessionStorage', () => {
setupTestDbHooks()
afterEach(() => {
vi.restoreAllMocks()
})
const mockModelConfig: PrismaJson.ModelConfig = {
reasoning: false,
input: ['text'],
@@ -392,7 +396,7 @@ describe('DatabaseSessionStorage', () => {
expect(entries[2].id).toBe('01C')
})
it('should use the current comment ID for media tools in subsequent turns', async () => {
it('should use the current comment ID for media tools in subsequent turns (image)', async () => {
const { agent, user, team } = await setupTestData()
// Create a project and asset to satisfy FK constraints
@@ -405,6 +409,8 @@ describe('DatabaseSessionStorage', () => {
name: 'test.png',
type: 'file',
mediaType: 'image/png',
// eslint-disable-next-line @typescript-eslint/no-explicit-any -- partial MediaInfo mock for testing
media: { proxyType: 'image' } as any,
projectId: project.id,
status: 'uploaded',
},
@@ -439,8 +445,62 @@ describe('DatabaseSessionStorage', () => {
providers: [],
})
// 4. Assert that the tools were instantiated using the new comment ID, not the original one
expect(createAnalyzeImageToolSpy).toHaveBeenCalledWith(asset.id, 'new-comment-456')
expect(createScreenshotToolSpy).toHaveBeenCalledWith(asset.id, 'new-comment-456')
// 4. Assert that the tools were instantiated using the new comment ID and user ID
expect(createAnalyzeImageToolSpy).toHaveBeenCalledWith(user.id, 'new-comment-456')
expect(createScreenshotToolSpy).toHaveBeenCalledWith(user.id, 'new-comment-456')
})
it('should use the current comment ID for media tools in subsequent turns (video)', async () => {
const { agent, user, team } = await setupTestData()
// Create a project and asset to satisfy FK constraints
const project = await prisma.project.create({
data: { name: 'Test Project', teamId: team.id },
})
const asset = await prisma.asset.create({
data: {
id: 'asset-456',
name: 'test.mp4',
type: 'file',
mediaType: 'video/mp4',
// eslint-disable-next-line @typescript-eslint/no-explicit-any -- partial MediaInfo mock for testing
media: { proxyType: 'video' } as any,
projectId: project.id,
status: 'uploaded',
},
})
// 1. Create a session with an original comment ID and an asset ID using the storage class
const initialStorage = await DatabaseSessionStorage.create({
agentId: agent.id,
userId: user.id,
cwd: '/test/cwd',
assetId: asset.id,
userCommentId: 'original-comment-123',
})
const sessionId = initialStorage.sessionId
// 2. Spy on media tools constructors
const createAnalyzeImageToolSpy = vi.spyOn(analyzeImageModule, 'createAnalyzeImageTool')
const createScreenshotToolSpy = vi.spyOn(screenshotModule, 'createScreenshotTool')
// 3. Resume the session via createAgentSession with a NEW comment ID
await createAgentSession({
teamId: 't1',
agentId: agent.id,
providerName: 'test-provider',
modelId: 'test-model',
systemPrompt: 'Test prompt',
teamSkills: [],
allowedDomains: [],
sessionId: sessionId,
userId: user.id,
userCommentId: 'new-comment-456',
providers: [],
})
// 4. Assert that the tools were instantiated using the new comment ID and user ID
expect(createScreenshotToolSpy).toHaveBeenCalledWith(user.id, 'new-comment-456')
expect(createAnalyzeImageToolSpy).toHaveBeenCalledWith(user.id, 'new-comment-456')
})
})
+82 -1
View File
@@ -1,5 +1,10 @@
import { describe, it, expect, vi } from 'vitest'
import { formatSkillsForPrompt, createAgentSession } from './index'
import {
formatSkillsForPrompt,
createAgentSession,
getModelFromDb,
type DbProviderInfo,
} from './index'
import { createSandboxedBashTool } from './tools/sandboxed-bash'
import { SandboxManager, type SandboxAskCallback } from '@anthropic-ai/sandbox-runtime'
import { prisma } from '@shumai/db'
@@ -43,6 +48,7 @@ describe('formatSkillsForPrompt', () => {
// Verify lines and XML elements
expect(result).toContain('<available_skills>')
expect(result).toContain('<skill>')
expect(result).toContain('<id>skill-1</id>')
expect(result).toContain('<name>Deploy &amp; Test</name>')
expect(result).toContain(
'<description>Deploys code &lt;deploy&gt; and &quot;runs&quot; tests.</description>',
@@ -329,4 +335,79 @@ describe('Sandbox Network isolation integration', () => {
initializeSpy.mockRestore()
}
})
it('should filter out denied tools from AgentHarness', async () => {
const team = await prisma.team.create({
data: { name: 'Test Team 3' },
})
await prisma.user.create({
data: {
id: 'test-agent-3',
name: 'Agent 3',
email: 'agent3@shumai.ai',
type: 'agent',
},
})
await prisma.agent.create({
data: {
id: 'test-agent-3',
teamId: team.id,
type: 'chat',
config: { provider: 'google', model: 'gemini', deniedTools: ['bash'] },
},
})
const initializeSpy = vi.spyOn(SandboxManager, 'initialize').mockResolvedValue()
try {
const { harness } = await createAgentSession({
teamId: team.id,
agentId: 'test-agent-3',
providerName: 'google',
modelId: 'gemini',
systemPrompt: 'prompt',
teamSkills: [],
allowedDomains: [],
providers: [],
})
const tools = harness.getTools()
const bashTool = tools.find((t) => t.name === 'bash')
expect(bashTool).toBeUndefined()
const readSkillTool = tools.find((t) => t.name === 'read_skill')
expect(readSkillTool).toBeDefined()
} finally {
initializeSpy.mockRestore()
}
})
})
describe('getModelFromDb', () => {
it('should correctly initialize non-built-in model with provider and id', () => {
const providers: DbProviderInfo[] = [
{
name: 'google',
config: { api: 'google-generative-ai', apiKey: 'GOOGLE_API_KEY' },
models: [
{
modelId: 'non-existent-or-future-model',
name: 'Future Model',
config: {
api: 'google-generative-ai',
reasoning: false,
input: ['text'],
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
contextWindow: 8192,
maxTokens: 4096,
},
},
],
},
]
const model = getModelFromDb(providers, 'google', 'non-existent-or-future-model')
expect(model).toBeDefined()
expect(model?.id).toBe('non-existent-or-future-model')
expect(model?.provider).toBe('google')
expect(model?.api).toBe('google-generative-ai')
})
})
+112 -89
View File
@@ -6,21 +6,95 @@ import {
type ThinkingLevel,
} from '@earendil-works/pi-agent-core'
import { NodeExecutionEnv } from '@earendil-works/pi-agent-core/node'
import { getModel } from '@earendil-works/pi-ai'
import { getModel } from '@earendil-works/pi-ai/compat'
import { agentService } from '@shumai/core/src/agent/agent'
import { prisma } from '@shumai/db'
import { Type, type TSchema } from '@sinclair/typebox'
import * as fs from 'fs'
import * as path from 'path'
import { prisma } from '@shumai/db'
import { agentService } from '@shumai/core/src/agent/agent'
import { DatabaseSessionStorage } from './database-session-storage'
import { createAnalyzeImageTool } from './tools/analyze-image'
import { createScreenshotTool } from './tools/screenshot'
import { createCreateFileTool } from './tools/create-file'
import { createCreateFolderTool } from './tools/create-folder'
import { createCreateVersionTool } from './tools/create-version'
import { createListAssetsTool } from './tools/list-assets'
import { createReadPdfPagesTool } from './tools/read-pdf-pages'
import { createReadSkillTool } from './tools/read-skill'
import { createSandboxedBashTool } from './tools/sandboxed-bash'
import { createScreenshotTool } from './tools/screenshot'
export interface DbModelInfo {
modelId: string
name: string | null
config: PrismaJson.ModelConfig
}
export interface DbProviderInfo {
name: string
config: PrismaJson.ProviderConfig
models: DbModelInfo[]
}
export function getModelFromDb(providers: DbProviderInfo[], providerName: string, modelId: string) {
const dbProvider = providers.find((p) => p.name === providerName)
const dbModel = dbProvider?.models.find((m) => m.modelId === modelId)
if (!dbProvider || !dbModel) return undefined
// Model contains complex internal types from pi-ai
// eslint-disable-next-line @typescript-eslint/no-explicit-any
let m: any
try {
// Try to get built-in model as template
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const builtIn = (getModel as any)(providerName, modelId)
if (!builtIn) {
throw new Error(`Model ${modelId} is not a built-in model`)
}
m = { ...builtIn }
} catch {
// Not a built-in model
m = {
id: modelId,
name: dbModel.name || modelId,
provider: providerName,
api: 'openai-responses',
baseUrl: '',
reasoning: false,
input: ['text'],
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
contextWindow: 4096,
maxTokens: 4096,
}
}
// Override with database config
if (dbProvider.config.baseUrl) {
m.baseUrl = dbProvider.config.baseUrl
}
if (dbProvider.config.api) {
m.api = dbProvider.config.api
}
Object.assign(m, dbModel.config)
if (dbModel.name) {
m.name = dbModel.name
}
return m
}
export function getApiKeyAndHeadersFromDb(providers: DbProviderInfo[], providerName: string) {
// Helper to resolve key from ENV if it matches an ENV var name
const resolveKey = (key: string | undefined) => (key ? process.env[key] || key : undefined)
// Look up the provider in our providers list
const dbProvider = providers.find((p) => p.name === providerName)
const providerApiKey = dbProvider?.config?.apiKey
if (providerApiKey) {
return { apiKey: resolveKey(providerApiKey)! }
}
return undefined
}
export interface CreateAgentSessionParams {
teamId: string
@@ -35,15 +109,7 @@ export interface CreateAgentSessionParams {
userCommentId?: string | null
customTools?: AgentTool[]
thinkingLevel?: string
providers: Array<{
name: string
config: PrismaJson.ProviderConfig
models: Array<{
modelId: string
name: string | null
config: PrismaJson.ModelConfig
}>
}>
providers: DbProviderInfo[]
}
export async function createAgentSession(params: CreateAgentSessionParams) {
@@ -73,52 +139,9 @@ export async function createAgentSession(params: CreateAgentSessionParams) {
const agentDir = path.join(process.cwd(), '.pi', 'agents', agentId)
if (!fs.existsSync(agentDir)) fs.mkdirSync(agentDir, { recursive: true })
// Helper to construct a Model object from database providers list
const getModelFromDb = (pName: string, mId: string) => {
const dbProvider = params.providers.find((p) => p.name === pName)
const dbModel = dbProvider?.models.find((m) => m.modelId === mId)
if (!dbProvider || !dbModel) return undefined
// Model contains complex internal types from pi-ai
// eslint-disable-next-line @typescript-eslint/no-explicit-any
let m: any
try {
// Try to get built-in model as template
// eslint-disable-next-line @typescript-eslint/no-explicit-any
m = { ...(getModel as any)(pName, mId) }
} catch {
// Not a built-in model
m = {
id: mId,
name: dbModel.name || mId,
provider: pName,
api: 'openai-responses',
baseUrl: '',
reasoning: false,
input: ['text'],
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
contextWindow: 4096,
maxTokens: 4096,
}
}
// Override with database config
if (dbProvider.config.baseUrl) {
m.baseUrl = dbProvider.config.baseUrl
}
if (dbProvider.config.api) {
m.api = dbProvider.config.api
}
Object.assign(m, dbModel.config)
if (dbModel.name) {
m.name = dbModel.name
}
return m
}
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const model = getModelFromDb(providerName, modelId) || (getModel as any)(providerName, modelId)
const model =
getModelFromDb(params.providers, providerName, modelId) ||
(getModel as unknown as (p: string, m: string) => unknown)(providerName, modelId)
const piDir = path.join(process.cwd(), '.pi')
if (!fs.existsSync(piDir)) fs.mkdirSync(piDir, { recursive: true })
@@ -206,15 +229,15 @@ export async function createAgentSession(params: CreateAgentSessionParams) {
}
const metadata = await storage.getMetadata()
const assetId = metadata.assetId
const userCommentId =
passedUserCommentId !== undefined ? passedUserCommentId : metadata.userCommentId
const mediaTools: AgentTool[] = []
if (assetId) {
if (userId) {
mediaTools.push(
createAnalyzeImageTool(assetId, userCommentId),
createScreenshotTool(assetId, userCommentId),
createAnalyzeImageTool(userId, userCommentId),
createScreenshotTool(userId, userCommentId),
createReadPdfPagesTool(userId, userCommentId),
)
}
@@ -224,7 +247,7 @@ export async function createAgentSession(params: CreateAgentSessionParams) {
sandboxState.blockedHost = ''
},
})
const readSkill = createReadSkillTool(onEnvsAdded)
const readSkill = createReadSkillTool(userId, onEnvsAdded)
const systemTools: AgentTool[] = []
if (userId) {
@@ -236,6 +259,15 @@ export async function createAgentSession(params: CreateAgentSessionParams) {
)
}
const agent = await prisma.agent.findUnique({
where: { id: agentId },
})
const agentConfig = agent?.config as PrismaJson.AgentConfig | null | undefined
const deniedTools = agentConfig?.deniedTools || []
const allTools = [...mediaTools, readSkill, sandboxedBash, ...systemTools, ...customTools]
const enabledTools = allTools.filter((tool) => !deniedTools.includes(tool.name))
const harness = new AgentHarness({
env: new NodeExecutionEnv({ cwd: process.cwd() }),
session,
@@ -261,19 +293,9 @@ export async function createAgentSession(params: CreateAgentSessionParams) {
return prompt
},
getApiKeyAndHeaders: async (targetModel) => {
// Helper to resolve key from ENV if it matches an ENV var name
const resolveKey = (key: string | undefined) => (key ? process.env[key] || key : undefined)
// Look up the provider in our providers list
const dbProvider = params.providers.find((p) => p.name === targetModel.provider)
const providerApiKey = dbProvider?.config?.apiKey
if (providerApiKey) {
return { apiKey: resolveKey(providerApiKey)! }
}
return undefined
return getApiKeyAndHeadersFromDb(params.providers, targetModel.provider)
},
tools: [...mediaTools, readSkill, sandboxedBash, ...systemTools, ...customTools],
tools: enabledTools,
})
return { session, harness }
@@ -346,6 +368,7 @@ export function formatSkillsForPrompt(
for (const skill of skills) {
const filePath = path.join(process.cwd(), '.pi', 'skills', skill.id, 'SKILL.md')
lines.push(' <skill>')
lines.push(` <id>${escapeXml(skill.id)}</id>`)
lines.push(` <name>${escapeXml(skill.name)}</name>`)
lines.push(` <description>${escapeXml(skill.description || 'No description')}</description>`)
lines.push(` <location>${escapeXml(filePath)}</location>`)
@@ -357,15 +380,15 @@ export function formatSkillsForPrompt(
return lines.join('\n')
}
import { registerWorkflow, registerActivities } from '@shumai/workflow-core'
import { WorkflowTaskType } from '@shumai/db'
import { agentEmbeddingMedia } from './workflows/agent-embedding'
import { queryEmbeddingForSearch } from './workflows/query-embedding-for-search'
import { agentAutofillMedia } from './workflows/agent-autofill'
import { agentChat } from './workflows/agent-chat'
import { agentToolCall } from './workflows/agent-tool-call'
import { registerActivities, registerWorkflow } from '@shumai/workflow-core'
import * as agentActivities from './activities/agent'
import * as aiActivities from './activities/ai'
import { agentAutofillMedia } from './workflows/agent-autofill'
import { agentChat } from './workflows/agent-chat'
import { agentEmbeddingMedia } from './workflows/agent-embedding'
import { agentToolCall } from './workflows/agent-tool-call'
import { queryEmbeddingForSearch } from './workflows/query-embedding-for-search'
export function initAgentWorkflows() {
registerWorkflow(WorkflowTaskType.ai_embedding, agentEmbeddingMedia)
@@ -378,19 +401,19 @@ export function initAgentWorkflows() {
registerActivities(aiActivities)
}
export * from './database-session-storage'
export * from './activities/agent'
export {
extractAiMetadataActivity,
generateEmbeddingActivity,
generateImageEmbeddingActivity,
generateTextEmbeddingActivity,
generateVideoChunkEmbeddingActivity,
type ExtractAiMetadataParams,
type GeneratedEmbedding,
type GenerateEmbeddingParams,
generateEmbeddingActivity,
type GenerateTextEmbeddingParams,
generateTextEmbeddingActivity,
type ExtractAiMetadataParams,
extractAiMetadataActivity,
generateImageEmbeddingActivity,
generateVideoChunkEmbeddingActivity,
} from './activities/ai'
export * from './database-session-storage'
export * from './workflows/agent-autofill'
export * from './workflows/agent-chat'
export * from './workflows/agent-embedding'
+31 -6
View File
@@ -5,16 +5,21 @@ import {
WorkflowTaskStatus,
type Asset,
type AssetComment,
type User,
type WorkflowTask,
} from '@shumai/db'
import { s3Service } from '@shumai/core/src/s3/s3'
import { workflowService } from '@shumai/workflow-core'
import { authzService } from '@shumai/core/src/authz/authz'
vi.mock('@shumai/db', async (importOriginal) => {
const actual = await importOriginal<typeof import('@shumai/db')>()
return {
...actual,
prisma: {
user: {
findUnique: vi.fn(),
},
asset: {
findUnique: vi.fn(),
},
@@ -40,12 +45,23 @@ vi.mock('@shumai/workflow-core', () => ({
},
}))
vi.mock('@shumai/core/src/authz/authz', () => ({
authzService: {
hasPermission: vi.fn(),
},
Permission: { Read: 'Read' },
ResourceType: { Asset: 'asset' },
}))
describe('analyzeImageTool', () => {
beforeEach(() => {
vi.clearAllMocks()
})
it('should fetch the image from S3 when comment has no annotations', async () => {
it('should check user permissions and fetch the image from S3 when comment has no annotations', async () => {
vi.mocked(prisma.user.findUnique).mockResolvedValue({ id: 'user-1' } as User)
vi.mocked(authzService.hasPermission).mockResolvedValue()
vi.mocked(prisma.asset.findUnique).mockResolvedValue({
id: 'asset-1',
projectId: 'project-1',
@@ -55,6 +71,7 @@ describe('analyzeImageTool', () => {
vi.mocked(prisma.assetComment.findUnique).mockResolvedValue({
id: 'comment-1',
assetId: 'asset-1',
annotation: null,
} as unknown as AssetComment)
@@ -63,8 +80,15 @@ describe('analyzeImageTool', () => {
contentType: 'image/png',
} as unknown as { buffer: Buffer; contentType: string })
const tool = createAnalyzeImageTool('asset-1', 'comment-1')
const result = await tool.execute('call-1', {})
const tool = createAnalyzeImageTool('user-1', 'comment-1')
const result = await tool.execute('call-1', { assetId: 'asset-1' })
expect(authzService.hasPermission).toHaveBeenCalledWith({
user: { id: 'user-1' },
permission: 'Read',
type: 'asset',
id: 'asset-1',
})
expect(result.content[0].type).toBe('image')
expect((result.content[0] as unknown as { data: string }).data).toBe(
@@ -87,6 +111,7 @@ describe('analyzeImageTool', () => {
vi.mocked(prisma.assetComment.findUnique).mockResolvedValue({
id: 'comment-1',
assetId: 'asset-1',
annotation: [
{
type: 'box',
@@ -114,14 +139,14 @@ describe('analyzeImageTool', () => {
contentType: 'image/webp',
} as unknown as { buffer: Buffer; contentType: string })
const tool = createAnalyzeImageTool('asset-1', 'comment-1')
const result = await tool.execute('call-1', {})
const tool = createAnalyzeImageTool('user-1', 'comment-1')
const result = await tool.execute('call-1', { assetId: 'asset-1' })
expect(prisma.workflowTask.create).toHaveBeenCalledWith({
data: {
assetId: 'asset-1',
projectId: 'project-1',
type: 'transcode',
type: 'transcode_image_annotation',
status: 'pending',
payload: {
projectId: 'project-1',
+94 -96
View File
@@ -1,14 +1,19 @@
import { Type } from '@sinclair/typebox'
import { type AgentTool } from '@earendil-works/pi-agent-core'
import { type ImageContent } from '@earendil-works/pi-ai'
import { prisma, WorkflowTaskType, WorkflowTaskStatus } from '@shumai/db'
import { prisma, WorkflowTaskType, WorkflowTaskStatus, type User } from '@shumai/db'
import { s3Service } from '@shumai/core/src/s3/s3'
import { workflowService } from '@shumai/workflow-core'
import { authzService, Permission, ResourceType } from '@shumai/core/src/authz/authz'
const analyzeImageSchema = Type.Object({})
const analyzeImageSchema = Type.Object({
assetId: Type.String({
description: 'The asset ID of the image to analyze. This parameter is required.',
}),
})
export function createAnalyzeImageTool(
assetId: string,
userId: string,
userCommentId?: string | null,
): AgentTool<typeof analyzeImageSchema> {
return {
@@ -17,111 +22,104 @@ export function createAnalyzeImageTool(
description:
'Retrieves and views the image content. Call this tool if you need to analyze the image.',
parameters: analyzeImageSchema,
execute: async () => {
try {
const asset = await prisma.asset.findUnique({
where: { id: assetId },
include: { storageKey: true },
execute: async (_toolCallId, params) => {
const assetId = params.assetId
if (!userId) {
throw new Error('User ID is required for authorization.')
}
await authzService.hasPermission({
user: { id: userId } as User,
permission: Permission.Read,
type: ResourceType.Asset,
id: assetId,
})
const asset = await prisma.asset.findUnique({
where: { id: assetId },
include: { storageKey: true },
})
if (!asset) {
throw new Error(`Asset with ID ${assetId} not found.`)
}
// Get default media key
let mediaKey = asset.storageKey?.key
if (asset.media) {
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const mediaInfo = asset.media as any
if (mediaInfo.imageTranscodes && mediaInfo.imageTranscodes.length > 0) {
mediaKey = mediaInfo.imageTranscodes[0].key || mediaKey
}
}
if (!mediaKey) {
throw new Error('No media content found for this asset.')
}
// Check if trigger comment has draw annotations for this asset
let annotations: unknown = null
if (userCommentId) {
const comment = await prisma.assetComment.findUnique({
where: { id: userCommentId },
})
if (!asset) {
return {
content: [{ type: 'text', text: `Asset with ID ${assetId} not found.` }],
details: {},
if (comment && comment.assetId === assetId && comment.annotation) {
const list = comment.annotation
if (Array.isArray(list) && list.length > 0) {
annotations = list
}
}
}
// Get default media key
let mediaKey = asset.storageKey?.key
if (asset.media) {
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const mediaInfo = asset.media as any
if (mediaInfo.imageTranscodes && mediaInfo.imageTranscodes.length > 0) {
mediaKey = mediaInfo.imageTranscodes[0].key || mediaKey
}
}
let keyToUse = mediaKey
if (!mediaKey) {
return {
content: [{ type: 'text', text: 'No media content found for this asset.' }],
details: {},
}
}
// Check if trigger comment has draw annotations
let annotations: unknown = null
if (userCommentId) {
const comment = await prisma.assetComment.findUnique({
where: { id: userCommentId },
})
if (comment?.annotation) {
const list = comment.annotation
if (Array.isArray(list) && list.length > 0) {
annotations = list
}
}
}
let keyToUse = mediaKey
if (annotations) {
// Trigger transcode workflow to draw annotation
const task = await prisma.workflowTask.create({
data: {
assetId,
if (annotations) {
// Trigger transcode workflow to draw annotation
const task = await prisma.workflowTask.create({
data: {
assetId,
projectId: asset.projectId || 'none',
type: WorkflowTaskType.transcode_image_annotation,
status: WorkflowTaskStatus.pending,
payload: {
projectId: asset.projectId || 'none',
type: WorkflowTaskType.transcode,
status: WorkflowTaskStatus.pending,
payload: {
projectId: asset.projectId || 'none',
imageAnnotation: {
annotations: annotations as PrismaJson.AnnotationList,
},
imageAnnotation: {
annotations: annotations as PrismaJson.AnnotationList,
},
},
})
const completedTask = await workflowService.executeWait(task)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const output = completedTask.output as any
if (output?.key) {
keyToUse = output.key
} else {
throw new Error('Image annotation transcode workflow failed to output S3 key.')
}
}
// Retrieve the image content from S3
const bucket = process.env.S3_BUCKET || 'shumai'
const { buffer, contentType } = await s3Service.getObject(bucket, keyToUse)
const content: ImageContent[] = [
{
type: 'image',
data: buffer.toString('base64'),
mimeType: contentType || 'image/webp',
},
]
})
return {
content,
details: {
sourceKeys: [keyToUse],
},
}
} catch (error) {
return {
content: [
{
type: 'text',
text: `Error analyzing image: ${
error instanceof Error ? error.message : String(error)
}`,
},
],
details: {},
const completedTask = await workflowService.executeWait(task)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const output = completedTask.output as any
if (output?.key) {
keyToUse = output.key
} else {
throw new Error('Image annotation transcode workflow failed to output S3 key.')
}
}
// Retrieve the image content from S3
const bucket = process.env.S3_BUCKET || 'shumai'
const { buffer, contentType } = await s3Service.getObject(bucket, keyToUse)
const content: ImageContent[] = [
{
type: 'image',
data: buffer.toString('base64'),
mimeType: contentType || 'image/webp',
},
]
return {
content,
details: {
sourceKeys: [keyToUse],
},
}
},
}
}
+27 -36
View File
@@ -34,6 +34,9 @@ function getMimeType(filePath: string): string {
'.mp4': 'video/mp4',
'.mov': 'video/quicktime',
'.pdf': 'application/pdf',
'.txt': 'text/plain',
'.md': 'text/markdown',
'.markdown': 'text/markdown',
}
return mimeTypes[ext] || 'application/octet-stream'
}
@@ -55,45 +58,33 @@ export function createCreateFileTool(userId: string): AgentTool<typeof createFil
description: 'Create a new file in a specified parent folder from a local file path.',
parameters: createFileSchema,
execute: async (_toolCallId, params) => {
try {
const absolutePath = path.resolve(process.cwd(), params.path)
if (!fs.existsSync(absolutePath)) {
throw new Error(`Local file not found at path: ${params.path}`)
}
const absolutePath = path.resolve(process.cwd(), params.path)
if (!fs.existsSync(absolutePath)) {
throw new Error(`Local file not found at path: ${params.path}`)
}
const fileSize = fs.statSync(absolutePath).size
const mimeType = getMimeType(absolutePath)
const fileSize = fs.statSync(absolutePath).size
const mimeType = getMimeType(absolutePath)
// Generate compliant S3 key matching normal file upload format
const s3Key = `files/${ulid()}/${sanitizeFilename(path.basename(absolutePath))}`
await s3Service.uploadFileToKey(absolutePath, s3Key, mimeType)
// Generate compliant S3 key matching normal file upload format
const s3Key = `files/${ulid()}/${sanitizeFilename(path.basename(absolutePath))}`
await s3Service.uploadFileToKey(absolutePath, s3Key, mimeType)
const result = await executeAgentToolWorkflow({
toolName: 'create_file',
args: {
parent: params.parent,
s3Key,
name: path.basename(absolutePath),
size: fileSize,
contentType: mimeType,
},
userId,
assetId: params.parent,
})
return {
content: [{ type: 'text', text: JSON.stringify(result, null, 2) }],
details: result,
}
} catch (error) {
return {
content: [
{
type: 'text',
text: `Error creating file: ${error instanceof Error ? error.message : String(error)}`,
},
],
details: {},
}
const result = await executeAgentToolWorkflow({
toolName: 'create_file',
args: {
parent: params.parent,
s3Key,
name: path.basename(absolutePath),
size: fileSize,
contentType: mimeType,
},
userId,
assetId: params.parent,
})
return {
content: [{ type: 'text', text: JSON.stringify(result, null, 2) }],
details: result,
}
},
}
+9 -21
View File
@@ -19,27 +19,15 @@ export function createCreateFolderTool(userId: string): AgentTool<typeof createF
description: 'Create a new folder in a specified parent folder.',
parameters: createFolderSchema,
execute: async (_toolCallId, params) => {
try {
const result = await executeAgentToolWorkflow({
toolName: 'create_folder',
args: params,
userId,
assetId: params.parent,
})
return {
content: [{ type: 'text', text: JSON.stringify(result, null, 2) }],
details: result,
}
} catch (error) {
return {
content: [
{
type: 'text',
text: `Error creating folder: ${error instanceof Error ? error.message : String(error)}`,
},
],
details: {},
}
const result = await executeAgentToolWorkflow({
toolName: 'create_folder',
args: params,
userId,
assetId: params.parent,
})
return {
content: [{ type: 'text', text: JSON.stringify(result, null, 2) }],
details: result,
}
},
}
+27 -36
View File
@@ -34,6 +34,9 @@ function getMimeType(filePath: string): string {
'.mp4': 'video/mp4',
'.mov': 'video/quicktime',
'.pdf': 'application/pdf',
'.txt': 'text/plain',
'.md': 'text/markdown',
'.markdown': 'text/markdown',
}
return mimeTypes[ext] || 'application/octet-stream'
}
@@ -56,45 +59,33 @@ export function createCreateVersionTool(userId: string): AgentTool<typeof create
description: 'Create a new version of an existing file from a local file path.',
parameters: createVersionSchema,
execute: async (_toolCallId, params) => {
try {
const absolutePath = path.resolve(process.cwd(), params.path)
if (!fs.existsSync(absolutePath)) {
throw new Error(`Local file not found at path: ${params.path}`)
}
const absolutePath = path.resolve(process.cwd(), params.path)
if (!fs.existsSync(absolutePath)) {
throw new Error(`Local file not found at path: ${params.path}`)
}
const fileSize = fs.statSync(absolutePath).size
const mimeType = getMimeType(absolutePath)
const fileSize = fs.statSync(absolutePath).size
const mimeType = getMimeType(absolutePath)
// Generate compliant S3 key matching normal file upload format
const s3Key = `files/${ulid()}/${sanitizeFilename(path.basename(absolutePath))}`
await s3Service.uploadFileToKey(absolutePath, s3Key, mimeType)
// Generate compliant S3 key matching normal file upload format
const s3Key = `files/${ulid()}/${sanitizeFilename(path.basename(absolutePath))}`
await s3Service.uploadFileToKey(absolutePath, s3Key, mimeType)
const result = await executeAgentToolWorkflow({
toolName: 'create_version',
args: {
parent: params.parent,
s3Key,
name: path.basename(absolutePath),
size: fileSize,
contentType: mimeType,
},
userId,
assetId: params.parent,
})
return {
content: [{ type: 'text', text: JSON.stringify(result, null, 2) }],
details: result,
}
} catch (error) {
return {
content: [
{
type: 'text',
text: `Error creating file version: ${error instanceof Error ? error.message : String(error)}`,
},
],
details: {},
}
const result = await executeAgentToolWorkflow({
toolName: 'create_version',
args: {
parent: params.parent,
s3Key,
name: path.basename(absolutePath),
size: fileSize,
contentType: mimeType,
},
userId,
assetId: params.parent,
})
return {
content: [{ type: 'text', text: JSON.stringify(result, null, 2) }],
details: result,
}
},
}
+9 -21
View File
@@ -34,27 +34,15 @@ export function createListAssetsTool(userId: string): AgentTool<typeof listAsset
description: 'List the assets (files and folders) inside a parent folder with pagination.',
parameters: listAssetsSchema,
execute: async (_toolCallId, params) => {
try {
const result = await executeAgentToolWorkflow({
toolName: 'list_assets',
args: params,
userId,
assetId: params.parent,
})
return {
content: [{ type: 'text', text: JSON.stringify(result, null, 2) }],
details: result,
}
} catch (error) {
return {
content: [
{
type: 'text',
text: `Error listing assets: ${error instanceof Error ? error.message : String(error)}`,
},
],
details: {},
}
const result = await executeAgentToolWorkflow({
toolName: 'list_assets',
args: params,
userId,
assetId: params.parent,
})
return {
content: [{ type: 'text', text: JSON.stringify(result, null, 2) }],
details: result,
}
},
}
@@ -0,0 +1,209 @@
import { describe, it, expect, vi, beforeEach } from 'vitest'
import { createReadPdfPagesTool } from './read-pdf-pages'
import {
prisma,
WorkflowTaskStatus,
type Asset,
type AssetComment,
type User,
type WorkflowTask,
} from '@shumai/db'
import { s3Service } from '@shumai/core/src/s3/s3'
import { workflowService } from '@shumai/workflow-core'
import { authzService } from '@shumai/core/src/authz/authz'
vi.mock('@shumai/db', async (importOriginal) => {
const actual = await importOriginal<typeof import('@shumai/db')>()
return {
...actual,
prisma: {
user: {
findUnique: vi.fn(),
},
asset: {
findUnique: vi.fn(),
},
assetComment: {
findUnique: vi.fn(),
},
workflowTask: {
create: vi.fn(),
},
},
}
})
vi.mock('@shumai/core/src/s3/s3', () => ({
s3Service: {
getObject: vi.fn(),
},
}))
vi.mock('@shumai/workflow-core', () => ({
workflowService: {
executeWait: vi.fn(),
},
}))
vi.mock('@shumai/core/src/authz/authz', () => ({
authzService: {
hasPermission: vi.fn(),
},
Permission: { Read: 'Read' },
ResourceType: { Asset: 'asset' },
}))
describe('readPdfPagesTool', () => {
beforeEach(() => {
vi.clearAllMocks()
vi.mocked(authzService.hasPermission).mockResolvedValue()
})
it('should trigger pdfPages transcode workflow and return image outputs', async () => {
vi.mocked(prisma.user.findUnique).mockResolvedValue({ id: 'user-1' } as User)
vi.mocked(prisma.asset.findUnique).mockResolvedValue({
id: 'asset-1',
projectId: 'project-1',
media: { proxyType: 'pdf' },
} as unknown as Asset)
vi.mocked(prisma.assetComment.findUnique).mockResolvedValue({
id: 'comment-1',
assetId: 'asset-1',
second: 2,
annotation: [
{
type: 'box',
color: '#ff0000',
points: [
[0, 0],
[1, 1],
],
},
],
} as unknown as AssetComment)
vi.mocked(prisma.workflowTask.create).mockResolvedValue({
id: 'task-1',
} as unknown as WorkflowTask)
vi.mocked(workflowService.executeWait).mockResolvedValue({
id: 'task-1',
status: WorkflowTaskStatus.completed,
output: {
pages: [
{ key: 'pdf_pages/page_1.webp', page: 1 },
{ key: 'pdf_pages/page_2.webp', page: 2 },
],
},
} as unknown as WorkflowTask)
vi.mocked(s3Service.getObject).mockImplementation(async (_bucket: string, key: string) => {
return {
buffer: Buffer.from(`fake-bytes-for-${key}`),
contentType: 'image/webp',
} as unknown as { buffer: Buffer; contentType: string }
})
const tool = createReadPdfPagesTool('user-1', 'comment-1')
const result = await tool.execute('call-1', { assetId: 'asset-1', start: 1, end: 2 })
expect(authzService.hasPermission).toHaveBeenCalledWith({
user: { id: 'user-1' },
permission: 'Read',
type: 'asset',
id: 'asset-1',
})
expect(prisma.workflowTask.create).toHaveBeenCalledWith({
data: {
assetId: 'asset-1',
projectId: 'project-1',
type: 'transcode_pdf_pages',
status: 'pending',
payload: {
projectId: 'project-1',
pdfPages: {
start: 1,
end: 2,
commentTimestamp: 2,
annotations: [
{
type: 'box',
color: '#ff0000',
points: [
[0, 0],
[1, 1],
],
},
],
},
},
},
})
expect(result.content.length).toBe(2)
expect(result.content[0].type).toBe('image')
expect((result.content[0] as unknown as { data: string }).data).toBe(
Buffer.from('fake-bytes-for-pdf_pages/page_1.webp').toString('base64'),
)
expect(result.details.sourceKeys).toEqual(['pdf_pages/page_1.webp', 'pdf_pages/page_2.webp'])
})
it('should throw error if start page is less than 1', async () => {
const tool = createReadPdfPagesTool('user-1')
await expect(tool.execute('call-1', { assetId: 'asset-1', start: 0, end: 5 })).rejects.toThrow(
'Invalid page range: start page (0) must be at least 1.',
)
})
it('should throw error if start page is greater than end page', async () => {
const tool = createReadPdfPagesTool('user-1')
await expect(tool.execute('call-1', { assetId: 'asset-1', start: 5, end: 2 })).rejects.toThrow(
'Invalid page range: start page (5) must be less than or equal to end page (2).',
)
})
it('should throw error if page range exceeds maximum limit of 20', async () => {
const tool = createReadPdfPagesTool('user-1')
await expect(tool.execute('call-1', { assetId: 'asset-1', start: 1, end: 25 })).rejects.toThrow(
'Page range (25 pages requested) exceeds the maximum limit of 20 pages per request.',
)
})
it('should throw error if asset not found', async () => {
vi.mocked(prisma.asset.findUnique).mockResolvedValue(null)
const tool = createReadPdfPagesTool('user-1')
await expect(tool.execute('call-1', { assetId: 'asset-1', start: 1, end: 3 })).rejects.toThrow(
'Asset with ID asset-1 not found.',
)
})
it('should throw error if proxyType is not pdf', async () => {
vi.mocked(prisma.asset.findUnique).mockResolvedValue({
id: 'asset-1',
projectId: 'project-1',
media: { proxyType: 'image' },
} as unknown as Asset)
const tool = createReadPdfPagesTool('user-1')
await expect(tool.execute('call-1', { assetId: 'asset-1', start: 1, end: 3 })).rejects.toThrow(
'Asset asset-1 is not a PDF or document.',
)
})
it('should throw error if proxyType is null/undefined (unprocessed)', async () => {
vi.mocked(prisma.asset.findUnique).mockResolvedValue({
id: 'asset-1',
projectId: 'project-1',
media: null,
} as unknown as Asset)
const tool = createReadPdfPagesTool('user-1')
await expect(tool.execute('call-1', { assetId: 'asset-1', start: 1, end: 3 })).rejects.toThrow(
'Asset asset-1 is not a PDF or document.',
)
})
})
+140
View File
@@ -0,0 +1,140 @@
import { Type } from '@sinclair/typebox'
import { type AgentTool } from '@earendil-works/pi-agent-core'
import { type ImageContent } from '@earendil-works/pi-ai'
import { prisma, WorkflowTaskType, WorkflowTaskStatus, type User } from '@shumai/db'
import { s3Service } from '@shumai/core/src/s3/s3'
import { workflowService } from '@shumai/workflow-core'
import { authzService, Permission, ResourceType } from '@shumai/core/src/authz/authz'
const readPdfPagesSchema = Type.Object({
assetId: Type.String({
description: 'Asset ID of the PDF document. This parameter is required.',
}),
start: Type.Number({ description: 'Start page number (1-based index)' }),
end: Type.Number({ description: 'End page number (1-based index)' }),
})
export function createReadPdfPagesTool(
userId: string,
userCommentId?: string | null,
): AgentTool<typeof readPdfPagesSchema> {
return {
name: 'read_pdf_pages',
label: 'Read PDF Pages',
description:
'Converts specific page range of a PDF document into 1080p WebP images for visual analysis. Maximum 20 pages allowed per call.',
parameters: readPdfPagesSchema,
execute: async (_toolCallId, params) => {
if (params.start < 1) {
throw new Error(`Invalid page range: start page (${params.start}) must be at least 1.`)
}
if (params.start > params.end) {
throw new Error(
`Invalid page range: start page (${params.start}) must be less than or equal to end page (${params.end}).`,
)
}
const MAX_PAGE_RANGE = 20
const requestedPages = params.end - params.start + 1
if (requestedPages > MAX_PAGE_RANGE) {
throw new Error(
`Page range (${requestedPages} pages requested) exceeds the maximum limit of ${MAX_PAGE_RANGE} pages per request.`,
)
}
const targetAssetId = params.assetId
if (!userId) {
throw new Error('User ID is required for authorization.')
}
await authzService.hasPermission({
user: { id: userId } as User,
permission: Permission.Read,
type: ResourceType.Asset,
id: targetAssetId,
})
const asset = await prisma.asset.findUnique({
where: { id: targetAssetId },
})
if (!asset) {
throw new Error(`Asset with ID ${targetAssetId} not found.`)
}
const proxyType = (asset.media as PrismaJson.MediaInfo | null)?.proxyType
if (proxyType !== 'pdf') {
throw new Error(`Asset ${targetAssetId} is not a PDF or document.`)
}
// Fetch trigger comment details for comment page and annotations
let commentTimestamp: number | null = null
let annotations: unknown = null
if (userCommentId) {
const comment = await prisma.assetComment.findUnique({
where: { id: userCommentId },
})
if (comment && comment.assetId === targetAssetId) {
commentTimestamp = comment.second !== null ? comment.second : null
if (comment.annotation) {
const list = comment.annotation
if (Array.isArray(list) && list.length > 0) {
annotations = list
}
}
}
}
// Trigger transcode workflow to render PDF pages
const task = await prisma.workflowTask.create({
data: {
assetId: targetAssetId,
projectId: asset.projectId || 'none',
type: WorkflowTaskType.transcode_pdf_pages,
status: WorkflowTaskStatus.pending,
payload: {
projectId: asset.projectId || 'none',
pdfPages: {
start: params.start,
end: params.end,
commentTimestamp,
annotations: annotations as PrismaJson.AnnotationList,
},
},
},
})
const completedTask = await workflowService.executeWait(task)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const output = completedTask.output as any
const pages = output?.pages as Array<{ key: string; page: number }> | undefined
if (!pages || pages.length === 0) {
throw new Error('No PDF page images were generated.')
}
const content: ImageContent[] = []
const sourceKeys: string[] = []
const bucket = process.env.S3_BUCKET || 'shumai'
for (const pageItem of pages) {
const { buffer, contentType } = await s3Service.getObject(bucket, pageItem.key)
content.push({
type: 'image',
data: buffer.toString('base64'),
mimeType: contentType || 'image/webp',
})
sourceKeys.push(pageItem.key)
}
return {
content,
details: {
sourceKeys,
},
}
},
}
}
+63 -16
View File
@@ -18,16 +18,11 @@ describe('readSkillTool', () => {
vi.clearAllMocks()
})
it('should return error if skill not found', async () => {
const readSkillTool = createReadSkillTool(() => {})
const result = await readSkillTool.execute(
'1',
{ skillId: 'non-existent' },
undefined,
undefined,
)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
expect((result.content[0] as any).text).toContain('not found')
it('should throw error if skill not found', async () => {
const readSkillTool = createReadSkillTool(undefined, () => {})
await expect(
readSkillTool.execute('1', { skillId: 'non-existent' }, undefined, undefined),
).rejects.toThrow('not found')
})
it('should read skill from cache if hash matches', async () => {
@@ -49,7 +44,7 @@ describe('readSkillTool', () => {
return ''
})
const readSkillTool = createReadSkillTool(() => {})
const readSkillTool = createReadSkillTool(undefined, () => {})
const result = await readSkillTool.execute('1', { skillId: skill.id }, undefined, undefined)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
@@ -101,7 +96,7 @@ describe('readSkillTool', () => {
contentType: 'application/zip',
})
const readSkillTool = createReadSkillTool(() => {})
const readSkillTool = createReadSkillTool(undefined, () => {})
const result = await readSkillTool.execute('1', { skillId: skill.id }, undefined, undefined)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
@@ -146,7 +141,7 @@ describe('readSkillTool', () => {
})
let capturedEnvs: Record<string, string> = {}
const readSkillTool = createReadSkillTool((envs) => {
const readSkillTool = createReadSkillTool(undefined, (envs) => {
capturedEnvs = { ...capturedEnvs, ...envs }
})
await readSkillTool.execute('1', { skillId: skill.id }, undefined, undefined)
@@ -181,7 +176,7 @@ describe('readSkillTool', () => {
})
let capturedEnvs: Record<string, string> = {}
const readSkillTool = createReadSkillTool((envs) => {
const readSkillTool = createReadSkillTool(undefined, (envs) => {
capturedEnvs = { ...capturedEnvs, ...envs }
})
await readSkillTool.execute('1', { skillId: skill.id }, undefined, undefined)
@@ -216,7 +211,7 @@ describe('readSkillTool', () => {
})
let capturedEnvs: Record<string, string> = {}
const readSkillTool = createReadSkillTool((envs) => {
const readSkillTool = createReadSkillTool(undefined, (envs) => {
capturedEnvs = { ...capturedEnvs, ...envs }
})
await readSkillTool.execute('1', { skillId: skill.id }, undefined, undefined)
@@ -251,7 +246,7 @@ describe('readSkillTool', () => {
})
let capturedEnvs: Record<string, string> = {}
const readSkillTool = createReadSkillTool((envs) => {
const readSkillTool = createReadSkillTool(undefined, (envs) => {
capturedEnvs = { ...capturedEnvs, ...envs }
})
await readSkillTool.execute('1', { skillId: skill.id }, undefined, undefined)
@@ -261,4 +256,56 @@ describe('readSkillTool', () => {
})
})
})
describe('permissions', () => {
it('should allow loading skill matching user role level or lower', async () => {
const team = await prisma.team.create({ data: { name: 'Permission Team' } })
const ownerUser = await prisma.user.create({
data: { name: 'Owner User', email: 'owner@test.com' },
})
const reviewerUser = await prisma.user.create({
data: { name: 'Reviewer User', email: 'reviewer@test.com' },
})
await prisma.teamMember.create({
data: { teamId: team.id, userId: ownerUser.id, role: 'owner' },
})
await prisma.teamMember.create({
data: { teamId: team.id, userId: reviewerUser.id, role: 'reviewer' },
})
const ownerSkill = await prisma.skill.create({
data: {
name: 'Owner Skill',
assetId: 'asset1',
hash: 'hash-owner',
teamId: team.id,
permission: 'owner',
},
})
vi.spyOn(fs, 'existsSync').mockReturnValue(true)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
vi.spyOn(fs, 'readFileSync').mockImplementation((path: any) => {
if (path.toString().endsWith('.hash')) return 'hash-owner'
if (path.toString().endsWith('SKILL.md')) return '# Owner Skill'
return ''
})
const ownerTool = createReadSkillTool(ownerUser.id, () => {})
const resultOwner = await ownerTool.execute(
'1',
{ skillId: ownerSkill.id },
undefined,
undefined,
)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
expect((resultOwner.content[0] as any).text).toBe('# Owner Skill')
const reviewerTool = createReadSkillTool(reviewerUser.id, () => {})
await expect(
reviewerTool.execute('1', { skillId: ownerSkill.id }, undefined, undefined),
).rejects.toThrow('Permission denied')
})
})
})
+43 -30
View File
@@ -7,7 +7,14 @@ const readSkillSchema = Type.Object({
skillId: Type.String({ description: 'The ID of the skill to read' }),
})
const ROLE_HIERARCHY: Record<string, number> = {
owner: 3,
editor: 2,
reviewer: 1,
}
export const createReadSkillTool = (
userId: string | undefined,
onEnvsAdded: (envs: Record<string, string>) => void,
): AgentTool<typeof readSkillSchema, { skillId: string }> => ({
name: 'read_skill',
@@ -15,41 +22,47 @@ export const createReadSkillTool = (
description: 'Downloads and reads an agent skill to understand how to use it.',
parameters: readSkillSchema,
execute: async (_toolCallId, params) => {
try {
const skill = await prisma.skill.findUnique({
where: { id: params.skillId },
})
const skill = await prisma.skill.findUnique({
where: { id: params.skillId },
})
if (!skill) {
return {
content: [{ type: 'text', text: `Skill with ID ${params.skillId} not found.` }],
details: { skillId: params.skillId },
}
}
if (!skill) {
throw new Error(`Skill with ID ${params.skillId} not found.`)
}
// Capture environment variables using agentService
const envs = await agentService.getSkillEnvs(params.skillId)
onEnvsAdded(envs)
const requiredLevel = ROLE_HIERARCHY[skill.permission] || 1
// Retrieve skill content using agentService
const skillMdContent = await agentService.getSkillContent(params.skillId)
return {
content: [{ type: 'text', text: skillMdContent }],
details: { skillId: params.skillId },
}
} catch (error) {
return {
content: [
{
type: 'text',
text: `Error reading skill ${params.skillId}: ${
error instanceof Error ? error.message : String(error)
}`,
if (userId) {
const member = await prisma.teamMember.findUnique({
where: {
teamIdUserId: {
teamId: skill.teamId,
userId,
},
],
details: { skillId: params.skillId },
},
})
const userLevel = member ? ROLE_HIERARCHY[member.role] || 0 : 0
if (userLevel < requiredLevel) {
throw new Error(
`Permission denied: Insufficient role to load skill "${skill.name}". Minimum required role is "${skill.permission}".`,
)
}
} else if (requiredLevel > 1) {
throw new Error(
`Permission denied: User context required to load skill "${skill.name}". Minimum required role is "${skill.permission}".`,
)
}
// Capture environment variables using agentService
const envs = await agentService.getSkillEnvs(params.skillId)
onEnvsAdded(envs)
// Retrieve skill content using agentService
const skillMdContent = await agentService.getSkillContent(params.skillId)
return {
content: [{ type: 'text', text: skillMdContent }],
details: { skillId: params.skillId },
}
},
})
+27 -3
View File
@@ -5,16 +5,21 @@ import {
WorkflowTaskStatus,
type Asset,
type AssetComment,
type User,
type WorkflowTask,
} from '@shumai/db'
import { s3Service } from '@shumai/core/src/s3/s3'
import { workflowService } from '@shumai/workflow-core'
import { authzService } from '@shumai/core/src/authz/authz'
vi.mock('@shumai/db', async (importOriginal) => {
const actual = await importOriginal<typeof import('@shumai/db')>()
return {
...actual,
prisma: {
user: {
findUnique: vi.fn(),
},
asset: {
findUnique: vi.fn(),
},
@@ -40,12 +45,23 @@ vi.mock('@shumai/workflow-core', () => ({
},
}))
vi.mock('@shumai/core/src/authz/authz', () => ({
authzService: {
hasPermission: vi.fn(),
},
Permission: { Read: 'Read' },
ResourceType: { Asset: 'asset' },
}))
describe('screenshotTool', () => {
beforeEach(() => {
vi.clearAllMocks()
})
it('should trigger screenshot transcode workflow and return image outputs', async () => {
vi.mocked(prisma.user.findUnique).mockResolvedValue({ id: 'user-1' } as User)
vi.mocked(authzService.hasPermission).mockResolvedValue()
vi.mocked(prisma.asset.findUnique).mockResolvedValue({
id: 'asset-1',
projectId: 'project-1',
@@ -53,6 +69,7 @@ describe('screenshotTool', () => {
vi.mocked(prisma.assetComment.findUnique).mockResolvedValue({
id: 'comment-1',
assetId: 'asset-1',
second: 5.0,
annotation: [
{
@@ -88,14 +105,21 @@ describe('screenshotTool', () => {
} as unknown as { buffer: Buffer; contentType: string }
})
const tool = createScreenshotTool('asset-1', 'comment-1')
const result = await tool.execute('call-1', { start: 0, end: 10, count: 2 })
const tool = createScreenshotTool('user-1', 'comment-1')
const result = await tool.execute('call-1', { assetId: 'asset-1', start: 0, end: 10, count: 2 })
expect(authzService.hasPermission).toHaveBeenCalledWith({
user: { id: 'user-1' },
permission: 'Read',
type: 'asset',
id: 'asset-1',
})
expect(prisma.workflowTask.create).toHaveBeenCalledWith({
data: {
assetId: 'asset-1',
projectId: 'project-1',
type: 'transcode',
type: 'transcode_screenshot',
status: 'pending',
payload: {
projectId: 'project-1',
+85 -88
View File
@@ -1,18 +1,22 @@
import { Type } from '@sinclair/typebox'
import { type AgentTool } from '@earendil-works/pi-agent-core'
import { type ImageContent } from '@earendil-works/pi-ai'
import { prisma, WorkflowTaskType, WorkflowTaskStatus } from '@shumai/db'
import { prisma, WorkflowTaskType, WorkflowTaskStatus, type User } from '@shumai/db'
import { s3Service } from '@shumai/core/src/s3/s3'
import { workflowService } from '@shumai/workflow-core'
import { authzService, Permission, ResourceType } from '@shumai/core/src/authz/authz'
const screenshotSchema = Type.Object({
assetId: Type.String({
description: 'The asset ID of the video asset. This parameter is required.',
}),
start: Type.Number({ description: 'Start time in seconds' }),
end: Type.Number({ description: 'End time in seconds' }),
count: Type.Number({ description: 'Number of screenshots to take' }),
})
export function createScreenshotTool(
assetId: string,
userId: string,
userCommentId?: string | null,
): AgentTool<typeof screenshotSchema> {
return {
@@ -22,102 +26,95 @@ export function createScreenshotTool(
'Takes screenshots for a video at specified interval. It returns the images to the agent context.',
parameters: screenshotSchema,
execute: async (_toolCallId, params) => {
try {
const asset = await prisma.asset.findUnique({
where: { id: assetId },
const assetId = params.assetId
if (!userId) {
throw new Error('User ID is required for authorization.')
}
await authzService.hasPermission({
user: { id: userId } as User,
permission: Permission.Read,
type: ResourceType.Asset,
id: assetId,
})
const asset = await prisma.asset.findUnique({
where: { id: assetId },
})
if (!asset) {
throw new Error(`Asset with ID ${assetId} not found.`)
}
// Fetch trigger comment details for timestamp and annotations
let commentTimestamp: number | null = null
let annotations: unknown = null
if (userCommentId) {
const comment = await prisma.assetComment.findUnique({
where: { id: userCommentId },
})
if (!asset) {
return {
content: [{ type: 'text', text: `Asset with ID ${assetId} not found.` }],
details: {},
}
}
// Fetch trigger comment details for timestamp and annotations
let commentTimestamp: number | null = null
let annotations: unknown = null
if (userCommentId) {
const comment = await prisma.assetComment.findUnique({
where: { id: userCommentId },
})
if (comment) {
commentTimestamp = comment.second !== null ? comment.second : null
if (comment.annotation) {
const list = comment.annotation
if (Array.isArray(list) && list.length > 0) {
annotations = list
}
if (comment && comment.assetId === assetId) {
commentTimestamp = comment.second !== null ? comment.second : null
if (comment.annotation) {
const list = comment.annotation
if (Array.isArray(list) && list.length > 0) {
annotations = list
}
}
}
}
// Trigger transcode workflow to take screenshots
const task = await prisma.workflowTask.create({
data: {
assetId,
// Trigger transcode workflow to take screenshots
const task = await prisma.workflowTask.create({
data: {
assetId,
projectId: asset.projectId || 'none',
type: WorkflowTaskType.transcode_screenshot,
status: WorkflowTaskStatus.pending,
payload: {
projectId: asset.projectId || 'none',
type: WorkflowTaskType.transcode,
status: WorkflowTaskStatus.pending,
payload: {
projectId: asset.projectId || 'none',
screenshot: {
start: params.start,
end: params.end,
count: params.count,
commentTimestamp,
annotations: annotations as PrismaJson.AnnotationList,
},
screenshot: {
start: params.start,
end: params.end,
count: params.count,
commentTimestamp,
annotations: annotations as PrismaJson.AnnotationList,
},
},
},
})
const completedTask = await workflowService.executeWait(task)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const output = completedTask.output as any
const screenshots = output?.screenshots as
| Array<{ key: string; timestamp: number }>
| undefined
if (!screenshots || screenshots.length === 0) {
throw new Error('No screenshots were generated.')
}
const content: ImageContent[] = []
const sourceKeys: string[] = []
const bucket = process.env.S3_BUCKET || 'shumai'
for (const shot of screenshots) {
const { buffer, contentType } = await s3Service.getObject(bucket, shot.key)
content.push({
type: 'image',
data: buffer.toString('base64'),
mimeType: contentType || 'image/webp',
})
sourceKeys.push(shot.key)
}
const completedTask = await workflowService.executeWait(task)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const output = completedTask.output as any
const screenshots = output?.screenshots as
| Array<{ key: string; timestamp: number }>
| undefined
if (!screenshots || screenshots.length === 0) {
return {
content: [{ type: 'text', text: 'No screenshots were generated.' }],
details: {},
}
}
const content: ImageContent[] = []
const sourceKeys: string[] = []
const bucket = process.env.S3_BUCKET || 'shumai'
for (const shot of screenshots) {
const { buffer, contentType } = await s3Service.getObject(bucket, shot.key)
content.push({
type: 'image',
data: buffer.toString('base64'),
mimeType: contentType || 'image/webp',
})
sourceKeys.push(shot.key)
}
return {
content,
details: {
sourceKeys,
},
}
} catch (error) {
return {
content: [
{
type: 'text',
text: `Error taking screenshots: ${
error instanceof Error ? error.message : String(error)
}`,
},
],
details: {},
}
return {
content,
details: {
sourceKeys,
},
}
},
}
@@ -64,6 +64,7 @@ describe('Agent Autofill Workflow', () => {
storageKey: { key: 'asset-key' },
project: { id: 'p1', teamId: 't1' },
mediaType: 'image/png',
media: { proxyType: 'image' },
})
mockActivities.downloadMediaToTmpActivity.mockResolvedValue({
filePath: '/tmp/test.png',
@@ -283,4 +284,33 @@ describe('Agent Autofill Workflow', () => {
tmpDir: '/tmp/test-dir',
})
})
it('should fail with non-retryable ApplicationFailure for unsupported audio files', async () => {
mockActivities.getAssetActivity.mockResolvedValue({
id: 'a1',
storageKey: { key: 'asset-key' },
project: { id: 'p1', teamId: 't1' },
mediaType: 'audio/mp3',
media: { proxyType: 'audio' },
})
const task = await prisma.workflowTask.create({
data: {
type: 'ai_metadata_autofill',
status: 'pending',
assetId: 'a1',
},
})
await expect(agentAutofillMedia(task)).rejects.toThrow(
'unsupported media type for autofill: audio/mp3',
)
// Verify task failed status
expect(mockActivities.updateTaskStatusActivity).toHaveBeenCalledWith({
taskId: task.id,
status: 'failed',
output: { error: 'unsupported media type for autofill: audio/mp3' },
})
})
})
+10 -1
View File
@@ -64,7 +64,16 @@ export async function agentAutofillMedia(task: WorkflowTask): Promise<void> {
const projectId = asset.project.id
// 2. Prepare Data (Images)
const isImage = asset.mediaType?.startsWith('image/') || false
const proxyType = (asset.media as PrismaJson.MediaInfo | null)?.proxyType
const isImage = proxyType === 'image'
const isVideo = proxyType === 'video'
if (!isImage && !isVideo) {
throw ApplicationFailure.create({
message: `unsupported media type for autofill: ${asset.mediaType}`,
nonRetryable: true,
})
}
const download = await executeActivity(transcodeWorkerQueue, downloadMediaToTmpActivity, {
assetKey: key,
@@ -25,7 +25,13 @@ describe('AgentChatPromptBuilder', () => {
})
it('should build prompt with image asset details', () => {
const builder = new AgentChatPromptBuilder('a1').withAssetDetails('test.png', 'image/png')
const builder = new AgentChatPromptBuilder('a1').withAssetDetails(
'test.png',
'image/png',
undefined,
undefined,
'image',
)
const result = builder.build()
expect(result).toContain('File Name: test.png')
@@ -39,6 +45,8 @@ describe('AgentChatPromptBuilder', () => {
'movie.mp4',
'video/mp4',
12.34,
undefined,
'video',
)
const result = builder.build()
@@ -49,6 +57,45 @@ describe('AgentChatPromptBuilder', () => {
expect(result).not.toContain("call the 'analyze_image' tool")
})
it('should build prompt with PDF asset details and total pages', () => {
const builder = new AgentChatPromptBuilder('a1').withAssetDetails(
'doc.pdf',
'application/pdf',
undefined,
15,
'pdf',
)
const result = builder.build()
expect(result).toContain('File Name: doc.pdf')
expect(result).toContain('File Type: pdf')
expect(result).toContain('Total Pages: 15')
expect(result).toContain("call the 'read_pdf_pages' tool")
expect(result).toContain('Maximum 20 pages allowed per call.')
expect(result).not.toContain("call the 'screenshot' tool")
expect(result).not.toContain("call the 'analyze_image' tool")
})
it('should build prompt with comment page for PDF asset', () => {
const builder = new AgentChatPromptBuilder('a1')
.withAssetDetails('doc.pdf', 'application/pdf', undefined, 10, 'pdf')
.withCommentTimestamp(3)
const result = builder.build()
expect(result).toContain('Comment Page: 3')
})
it('should build prompt for PDF continuation mode with comment page', () => {
const builder = new AgentChatPromptBuilder('a1')
.withContinuation(true)
.withAssetDetails('doc.pdf', 'application/pdf', undefined, 10, 'pdf')
.withCommentTimestamp(4)
const result = builder.build()
expect(result).toContain('Comment Page: 4')
expect(result).not.toContain('Comment Timestamp:')
})
it('should build prompt with comment timestamp', () => {
const builder = new AgentChatPromptBuilder('a1').withCommentTimestamp(5.5)
const result = builder.build()
@@ -60,7 +107,7 @@ describe('AgentChatPromptBuilder', () => {
const builder = new AgentChatPromptBuilder('a1')
.withExplicitMention(true)
.withPathContext('src/main.ts')
.withAssetDetails('movie.mp4', 'video/mp4', 12.34)
.withAssetDetails('movie.mp4', 'video/mp4', 12.34, undefined, 'video')
.withCommentTimestamp(8.2)
const result = builder.build()
@@ -73,4 +120,48 @@ describe('AgentChatPromptBuilder', () => {
expect(result).toContain("call the 'screenshot' tool")
expect(result).toContain('The user explicitly mentioned you')
})
it('should build prompt with attached files and referenced assets', () => {
const builder = new AgentChatPromptBuilder('a1')
.withAttachedFiles(['- file1.txt', '- file2.png'])
.withReferencedAssets(['- folder1', '- file3.mp4'])
const result = builder.build()
expect(result).toContain('[Context: Attached Files & Referenced Assets]')
expect(result).toContain('Attached Files:\n- file1.txt\n- file2.png')
expect(result).toContain('Referenced Workspace Assets:\n- folder1\n- file3.mp4')
})
it('should build prompt for continuation mode returning only turn-specific modifications', () => {
const builder = new AgentChatPromptBuilder('a1')
.withContinuation(true)
.withAssetDetails('movie.mp4', 'video/mp4', 12.34)
.withPathContext('src/main.ts')
.withCommentTimestamp(8.2)
.withAttachedFiles(['- file1.txt'])
.withReferencedAssets(['- folder1'])
const result = builder.build()
expect(result).not.toContain('The user is discussing an asset with ID:')
expect(result).not.toContain('File Name: movie.mp4')
expect(result).not.toContain('Asset Path Context:')
expect(result).toContain('Comment Timestamp: 8.20 seconds')
expect(result).toContain('[Context: New Attached Files & Referenced Assets]')
expect(result).toContain('Attached Files:\n- file1.txt')
expect(result).toContain('Referenced Workspace Assets:\n- folder1')
})
it('should include location change context block in continuation mode when asset has changed', () => {
const builder = new AgentChatPromptBuilder('a2')
.withContinuation(true)
.withAssetChanged(true)
.withAssetDetails('new-folder', 'folder', undefined, undefined, 'folder')
.withPathContext('projects/my-proj/new-folder')
const result = builder.build()
expect(result).toContain('[Context: User Navigated to a New Location]')
expect(result).toContain('New Location Asset ID: a2')
expect(result).toContain('File Name: new-folder')
expect(result).toContain('Asset Path Context:\nprojects/my-proj/new-folder')
})
})
@@ -1,16 +1,7 @@
export function getSimpleMediaType(mediaType?: string): 'image' | 'video' | 'other' {
if (!mediaType) return 'other'
const lower = mediaType.toLowerCase()
if (
lower.startsWith('image/') ||
lower === 'application/x-photoshop' ||
lower === 'image/vnd.adobe.photoshop'
) {
return 'image'
}
if (lower.startsWith('video/')) {
return 'video'
}
export function getSimpleMediaType(proxyType?: string | null): 'image' | 'video' | 'pdf' | 'other' {
if (proxyType === 'image') return 'image'
if (proxyType === 'video') return 'video'
if (proxyType === 'pdf') return 'pdf'
return 'other'
}
@@ -18,24 +9,63 @@ export class AgentChatPromptBuilder {
private assetId: string
private assetName?: string
private mediaType?: string
private proxyType?: string
private videoDuration?: number
private totalPages?: number
private commentTimestamp?: number
private pathContext?: string
private explicitMention = false
private attachedFiles: string[] = []
private referencedAssets: string[] = []
private isContinuation = false
private hasAssetChanged = false
constructor(assetId: string) {
this.assetId = assetId
}
withContinuation(isContinuation: boolean): this {
this.isContinuation = isContinuation
return this
}
withAssetChanged(hasAssetChanged?: boolean): this {
this.hasAssetChanged = !!hasAssetChanged
return this
}
withAttachedFiles(files?: string[]): this {
if (files) {
this.attachedFiles = files
}
return this
}
withReferencedAssets(assets?: string[]): this {
if (assets) {
this.referencedAssets = assets
}
return this
}
withPathContext(pathContext?: string): this {
this.pathContext = pathContext
return this
}
withAssetDetails(name: string, mediaType: string | null, duration?: number): this {
withAssetDetails(
name: string,
mediaType: string | null,
duration?: number,
totalPages?: number,
proxyType?: string | null,
): this {
this.assetName = name
this.mediaType = mediaType || undefined
this.proxyType = proxyType || undefined
this.videoDuration = duration
this.totalPages = totalPages
return this
}
@@ -52,30 +82,93 @@ export class AgentChatPromptBuilder {
}
build(): string {
const effectiveType = this.proxyType
if (this.isContinuation) {
let instruction = ''
if (this.hasAssetChanged) {
instruction += `[Context: User Navigated to a New Location]\nNew Location Asset ID: ${this.assetId}`
if (this.assetName) {
instruction += `\nFile Name: ${this.assetName}`
}
if (effectiveType) {
const type = getSimpleMediaType(effectiveType)
instruction += `\nFile Type: ${type}`
if (type === 'video' && this.videoDuration !== undefined) {
instruction += `\nVideo Length: ${this.videoDuration.toFixed(2)} seconds`
} else if (type === 'pdf' && this.totalPages !== undefined) {
instruction += `\nTotal Pages: ${this.totalPages}`
}
}
if (this.pathContext) {
instruction += `\n\nAsset Path Context:\n${this.pathContext}`
}
}
if (this.commentTimestamp !== undefined) {
const type = getSimpleMediaType(effectiveType)
if (instruction) instruction += '\n\n'
if (type === 'pdf') {
instruction += `Comment Page: ${Math.round(this.commentTimestamp)}`
} else {
instruction += `Comment Timestamp: ${this.commentTimestamp.toFixed(2)} seconds`
}
}
if (this.attachedFiles.length > 0 || this.referencedAssets.length > 0) {
if (instruction) instruction += '\n\n'
instruction += `[Context: New Attached Files & Referenced Assets]`
if (this.attachedFiles.length > 0) {
instruction += `\nAttached Files:\n${this.attachedFiles.join('\n')}`
}
if (this.referencedAssets.length > 0) {
instruction += `\nReferenced Workspace Assets:\n${this.referencedAssets.join('\n')}`
}
}
return instruction.trim()
}
let instruction = `The user is discussing an asset with ID: ${this.assetId}.`
if (this.assetName) {
instruction += `\nFile Name: ${this.assetName}`
}
if (this.mediaType) {
const type = getSimpleMediaType(this.mediaType)
if (effectiveType) {
const type = getSimpleMediaType(effectiveType)
instruction += `\nFile Type: ${type}`
if (type === 'video' && this.videoDuration !== undefined) {
instruction += `\nVideo Length: ${this.videoDuration.toFixed(2)} seconds`
} else if (type === 'pdf' && this.totalPages !== undefined) {
instruction += `\nTotal Pages: ${this.totalPages}`
}
}
if (this.commentTimestamp !== undefined) {
instruction += `\nComment Timestamp: ${this.commentTimestamp.toFixed(2)} seconds`
const type = getSimpleMediaType(effectiveType)
if (type === 'pdf') {
instruction += `\nComment Page: ${Math.round(this.commentTimestamp)}`
} else {
instruction += `\nComment Timestamp: ${this.commentTimestamp.toFixed(2)} seconds`
}
}
if (this.pathContext) {
instruction += `\n\nAsset Path Context:\n${this.pathContext}`
}
if (this.mediaType) {
const type = getSimpleMediaType(this.mediaType)
if (this.attachedFiles.length > 0 || this.referencedAssets.length > 0) {
instruction += `\n\n[Context: Attached Files & Referenced Assets]`
if (this.attachedFiles.length > 0) {
instruction += `\nAttached Files:\n${this.attachedFiles.join('\n')}`
}
if (this.referencedAssets.length > 0) {
instruction += `\nReferenced Workspace Assets:\n${this.referencedAssets.join('\n')}`
}
}
if (effectiveType) {
const type = getSimpleMediaType(effectiveType)
if (type === 'image') {
instruction += `\n\nIf you need to view the image data, call the 'analyze_image' tool. It does not require any parameters.`
} else if (type === 'video') {
instruction += `\n\nIf you need to view visual frames or take screenshots of the video, call the 'screenshot' tool. You must specify the 'start' (seconds), 'end' (seconds), and 'count' (number of screenshots) parameters.`
} else if (type === 'pdf') {
instruction += `\n\nIf you need to view pages of the PDF document, call the 'read_pdf_pages' tool. You must specify the 'start' (page number) and 'end' (page number) parameters. Maximum 20 pages allowed per call.`
}
}
+173 -3
View File
@@ -46,17 +46,22 @@ describe('Agent Chat Workflow', () => {
getAssetPathContextActivity: Object.assign(vi.fn(), {
_activityName: 'getAssetPathContextActivity',
}),
generateSessionNameActivity: Object.assign(vi.fn(), {
_activityName: 'generateSessionNameActivity',
}),
}
mockActivities.getAgentWorkerQueueActivity.mockResolvedValue('agent_queue')
mockActivities.updateTaskStatusActivity.mockResolvedValue({})
mockActivities.createCommentActivity.mockResolvedValue({ id: 'comment-placeholder-id' })
mockActivities.generateSessionNameActivity.mockResolvedValue(undefined)
mockActivities.getAssetActivity.mockResolvedValue({
id: 'a1',
project: { teamId: 't1' },
type: 'file',
name: 'test-file.png',
mediaType: 'image/png',
media: { proxyType: 'image' },
parentId: 'parent-folder-id',
})
mockActivities.getCommentActivity.mockResolvedValue({
@@ -129,7 +134,7 @@ describe('Agent Chat Workflow', () => {
// Verify agent instruction composition
const expectedInstruction1 = new AgentChatPromptBuilder('a1')
.withPathContext('Path: folder/subfolder/file.png')
.withAssetDetails('test-file.png', 'image/png', undefined)
.withAssetDetails('test-file.png', 'image/png', undefined, undefined, 'image')
.withCommentTimestamp(null)
.withExplicitMention(true)
.build()
@@ -166,6 +171,15 @@ describe('Agent Chat Workflow', () => {
status: 'completed',
output: { sessionId: 'session-123' },
})
// Verify session name generation triggered
expect(mockActivities.generateSessionNameActivity).toHaveBeenCalledWith({
teamId: 't1',
agentId: 'b1',
prompt: 'hello agent',
sessionId: 'session-123',
context: expect.any(Object),
})
})
it('should skip session initialization if sessionId is already provided', async () => {
@@ -185,6 +199,7 @@ describe('Agent Chat Workflow', () => {
expect(mockActivities.initializeAgentSessionActivity).not.toHaveBeenCalled()
const expectedInstruction2 = new AgentChatPromptBuilder('a1')
.withContinuation(true)
.withPathContext('Path: folder/subfolder/file.png')
.withAssetDetails('test-file.png', 'image/png', undefined)
.withCommentTimestamp(null)
@@ -197,6 +212,9 @@ describe('Agent Chat Workflow', () => {
agentsInstruction: expectedInstruction2,
}),
)
// Verify session name generation is skipped
expect(mockActivities.generateSessionNameActivity).not.toHaveBeenCalled()
})
it('should delete placeholder comment and finish when AI responds with __NO_REPLY__', async () => {
@@ -222,7 +240,7 @@ describe('Agent Chat Workflow', () => {
// Verify instruction contains not explicitly mentioned instructions
const expectedInstruction3 = new AgentChatPromptBuilder('a1')
.withPathContext('Path: folder/subfolder/file.png')
.withAssetDetails('test-file.png', 'image/png', undefined)
.withAssetDetails('test-file.png', 'image/png', undefined, undefined, 'image')
.withCommentTimestamp(null)
.withExplicitMention(false)
.build()
@@ -254,18 +272,21 @@ describe('Agent Chat Workflow', () => {
{
asset: {
mediaType: 'image/png',
media: { proxyType: 'image' },
storageKey: { key: 'attachments/image1.png' },
},
},
{
asset: {
mediaType: 'video/mp4', // non-image, should be ignored
media: { proxyType: 'video' },
storageKey: { key: 'attachments/video.mp4' },
},
},
{
asset: {
mediaType: 'image/jpeg',
media: { proxyType: 'image' },
storageKey: { key: 'attachments/image2.jpg' },
},
},
@@ -288,7 +309,7 @@ describe('Agent Chat Workflow', () => {
const expectedInstruction4 = new AgentChatPromptBuilder('a1')
.withPathContext('Path: folder/subfolder/file.png')
.withAssetDetails('test-file.png', 'image/png', undefined)
.withAssetDetails('test-file.png', 'image/png', undefined, undefined, 'image')
.withCommentTimestamp(null)
.withExplicitMention(false)
.build()
@@ -397,4 +418,153 @@ describe('Agent Chat Workflow', () => {
message: 'Sorry, I encountered an error while processing your request.',
})
})
it('should execute direct-context chatbot flow resolving file metadata and skipping comment placeholder', async () => {
mockActivities.getAssetActivity.mockImplementation(async (id: string) => {
if (id === 'file-attachment-1') {
return {
id: 'file-attachment-1',
name: 'attachment.png',
type: 'file',
mediaType: 'image/png',
projectId: 'p1',
}
}
if (id === 'referenced-asset-1') {
return {
id: 'referenced-asset-1',
name: 'ref-folder',
type: 'folder',
mediaType: null,
projectId: 'p1',
}
}
return {
id,
name: 'test-file.png',
type: 'file',
mediaType: 'image/png',
projectId: 'p1',
project: { teamId: 't1' },
}
})
mockActivities.getAssetPathContextActivity.mockImplementation(async (id: string) => {
if (id === 'file-attachment-1') return 'attachment.png'
if (id === 'referenced-asset-1') return 'ref-folder'
return 'folder/subfolder/file.png'
})
const task = await prisma.workflowTask.create({
data: {
type: 'chat',
status: 'pending',
assetId: 'a1',
payload: {
projectId: 'p1',
agent: {
agentId: 'b1',
sessionId: 'session-direct-123',
prompt: 'chatbot prompt',
attachedFiles: ['file-attachment-1'],
assetIds: ['referenced-asset-1'],
explicitMention: true,
},
},
},
})
await agentChat(task)
// Verify placeholder comment is NOT created
expect(mockActivities.createCommentActivity).not.toHaveBeenCalled()
expect(mockActivities.updateCommentActivity).not.toHaveBeenCalled()
// Verify prompt builder is populated and workflow runs agentChatActivity with context instruction
const expectedInstruction = new AgentChatPromptBuilder('a1')
.withContinuation(true)
.withPathContext('folder/subfolder/file.png')
.withAssetDetails('test-file.png', 'image/png', undefined)
.withCommentTimestamp(undefined)
.withExplicitMention(true)
.withAttachedFiles([
'- Name: attachment.png (ID: file-attachment-1, Type: file, Media Type: image/png, Project ID: p1, Path: attachment.png)',
])
.withReferencedAssets([
'- Name: ref-folder (ID: referenced-asset-1, Type: folder, Media Type: unknown, Project ID: p1, Path: ref-folder)',
])
.build()
expect(mockActivities.agentChatActivity).toHaveBeenCalledWith(
expect.objectContaining({
teamId: 't1',
agentId: 'b1',
message: 'chatbot prompt',
imageUrls: [],
projectId: 'p1',
folderId: '',
agentsInstruction: expectedInstruction,
sessionId: 'session-direct-123',
userId: undefined,
userCommentId: undefined,
explicitMention: true,
context: { agent: { id: 'b1' } },
attachedAssets: [
{ id: 'file-attachment-1', name: 'attachment.png', type: 'file' },
{ id: 'referenced-asset-1', name: 'ref-folder', type: 'folder' },
],
}),
)
})
it('should handle version stack asset by building instructions with resolved latest version asset details', async () => {
mockActivities.getAssetActivity.mockImplementation(async (id: string) => {
if (id === 'vs-1') {
return {
id: 'v2-latest',
project: { teamId: 't1' },
type: 'file',
name: 'latest-video.mp4',
mediaType: 'video/mp4',
media: { proxyType: 'video', duration: 42 },
parentId: 'vs-1',
}
}
return null
})
const task = await prisma.workflowTask.create({
data: {
type: 'chat',
status: 'pending',
assetId: 'vs-1',
payload: {
projectId: 'p1',
agent: {
prompt: 'Explain this version stack video',
sessionId: 'session-vs-123',
agentId: 'b1',
},
},
},
})
await agentChat(task)
const expectedInstruction = new AgentChatPromptBuilder('v2-latest')
.withContinuation(true)
.withPathContext('Path: folder/subfolder/file.png')
.withAssetDetails('latest-video.mp4', 'video/mp4', 42, undefined, 'video')
.build()
expect(mockActivities.agentChatActivity).toHaveBeenCalledWith(
expect.objectContaining({
teamId: 't1',
agentId: 'b1',
message: 'Explain this version stack video',
agentsInstruction: expectedInstruction,
sessionId: 'session-vs-123',
}),
)
})
})
+123 -37
View File
@@ -17,6 +17,7 @@ export async function agentChat(task: WorkflowTask): Promise<void> {
deleteCommentActivity,
initializeAgentSessionActivity,
getAssetPathContextActivity,
generateSessionNameActivity,
} = getActivities()
let placeholderCommentId: string | undefined
@@ -46,26 +47,43 @@ export async function agentChat(task: WorkflowTask): Promise<void> {
// 1. Get User Comment
const userCommentId = payload.agent?.userCommentId
if (!userCommentId) {
throw ApplicationFailure.create({
message: 'userCommentId missing in payload',
nonRetryable: true,
})
}
const userComment = await executeActivity(agentWorkerQueue, getCommentActivity, userCommentId)
if (!userComment) {
throw ApplicationFailure.create({ message: 'User comment not found', nonRetryable: true })
}
let prompt = ''
let attachmentImageUrls: string[] = []
let commentTimestamp: number | undefined
// 2. Create Placeholder Comment
const placeholder = await executeActivity(agentWorkerQueue, createCommentActivity, {
assetId: task.assetId,
message: '__CHAT__',
sessionId: sessionId || 'pending',
agentId: agentId,
replyToId: userComment.replyToId ?? userComment.id,
})
placeholderCommentId = placeholder.id
if (userCommentId) {
const userComment = await executeActivity(agentWorkerQueue, getCommentActivity, userCommentId)
if (!userComment) {
throw ApplicationFailure.create({ message: 'User comment not found', nonRetryable: true })
}
prompt = userComment.message || ''
if (userComment.second !== null && userComment.second !== undefined) {
commentTimestamp = userComment.second
}
// 2. Create Placeholder Comment
const placeholder = await executeActivity(agentWorkerQueue, createCommentActivity, {
assetId: task.assetId,
message: '__CHAT__',
sessionId: sessionId || 'pending',
agentId: agentId,
replyToId: userComment.replyToId ?? userComment.id,
})
placeholderCommentId = placeholder.id
// 5. Prepare Images (Attachments only)
if (userComment.attachments) {
for (const att of userComment.attachments) {
const attProxyType = (att.asset?.media as PrismaJson.MediaInfo | null)?.proxyType
if (attProxyType === 'image' && att.asset?.storageKey?.key) {
attachmentImageUrls.push(att.asset.storageKey.key)
}
}
}
} else {
prompt = payload.agent?.prompt || ''
attachmentImageUrls = payload.agent?.imageUrls || []
}
// 3. Get Asset
const asset = await executeActivity(agentWorkerQueue, getAssetActivity, task.assetId)
@@ -75,13 +93,21 @@ export async function agentChat(task: WorkflowTask): Promise<void> {
const teamId = asset.project.teamId
// 4. Initialize Session if missing
let isNewChat = !payload.agent?.sessionId || payload.agent?.isNewChat === true
if (!sessionId) {
if (!userCommentId) {
throw ApplicationFailure.create({
message: 'sessionId is required when userCommentId is missing',
nonRetryable: true,
})
}
sessionId = await executeActivity(agentWorkerQueue, initializeAgentSessionActivity, {
teamId,
agentId: agentId,
userCommentId,
userId: payload.agent?.userId,
})
isNewChat = true
}
// 4b. Fetch Agent Context
@@ -90,16 +116,6 @@ export async function agentChat(task: WorkflowTask): Promise<void> {
agentId: agentId,
})
// 5. Prepare Images (Attachments only)
const attachmentImageUrls: string[] = []
if (userComment?.attachments) {
for (const att of userComment.attachments) {
if (att.asset?.mediaType?.startsWith('image/') && att.asset?.storageKey?.key) {
attachmentImageUrls.push(att.asset.storageKey.key)
}
}
}
const pathContext = await executeActivity(
agentWorkerQueue,
getAssetPathContextActivity,
@@ -109,12 +125,53 @@ export async function agentChat(task: WorkflowTask): Promise<void> {
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const mediaInfo = asset.media as any
const duration = mediaInfo?.duration
const totalPages = mediaInfo?.frames || mediaInfo?.metadata?.totalFrames
// Resolve Attached Files Details
const attachedFileDetailsList: string[] = []
const attachedAssets: Array<{ id: string; name: string; type: string }> = []
if (payload.agent?.attachedFiles) {
for (const fileId of payload.agent.attachedFiles) {
const file = await executeActivity(agentWorkerQueue, getAssetActivity, fileId)
if (file) {
const path = await executeActivity(agentWorkerQueue, getAssetPathContextActivity, fileId)
attachedFileDetailsList.push(
`- Name: ${file.name} (ID: ${file.id}, Type: ${file.type}, Media Type: ${file.mediaType || 'unknown'}, Project ID: ${file.projectId || 'unknown'}, Path: ${path})`,
)
attachedAssets.push({ id: file.id, name: file.name, type: file.type })
}
}
}
// Resolve Referenced Workspace Assets Details
const referencedAssetDetailsList: string[] = []
if (payload.agent?.assetIds) {
for (const assetId of payload.agent.assetIds) {
const referencedAsset = await executeActivity(agentWorkerQueue, getAssetActivity, assetId)
if (referencedAsset) {
const path = await executeActivity(agentWorkerQueue, getAssetPathContextActivity, assetId)
referencedAssetDetailsList.push(
`- Name: ${referencedAsset.name} (ID: ${referencedAsset.id}, Type: ${referencedAsset.type}, Media Type: ${referencedAsset.mediaType || 'unknown'}, Project ID: ${referencedAsset.projectId || 'unknown'}, Path: ${path})`,
)
attachedAssets.push({
id: referencedAsset.id,
name: referencedAsset.name,
type: referencedAsset.type,
})
}
}
}
const proxyType = (asset.media as PrismaJson.MediaInfo | null)?.proxyType
const instruction = new AgentChatPromptBuilder(asset.id)
.withContinuation(!isNewChat)
.withAssetChanged(payload.agent?.hasAssetChanged)
.withPathContext(pathContext)
.withAssetDetails(asset.name, asset.mediaType, duration)
.withCommentTimestamp(userComment.second)
.withAssetDetails(asset.name, asset.mediaType, duration, totalPages, proxyType)
.withCommentTimestamp(commentTimestamp)
.withExplicitMention(payload.agent?.explicitMention)
.withAttachedFiles(attachedFileDetailsList)
.withReferencedAssets(referencedAssetDetailsList)
.build()
// 6. Call AI Chat
@@ -126,20 +183,34 @@ export async function agentChat(task: WorkflowTask): Promise<void> {
}
const aiResult = await executeActivity(agentWorkerQueue, agentChatActivity, {
taskId: task.id,
teamId,
agentId: agentId,
message: userComment.message || '',
message: prompt,
imageUrls: attachmentImageUrls,
projectId: payload.projectId,
folderId,
agentsInstruction: instruction,
sessionId,
userId: payload.agent?.userId,
userCommentId,
userCommentId: userCommentId || undefined,
explicitMention: payload.agent?.explicitMention,
context,
attachedAssets,
})
if (isNewChat) {
await executeActivity(agentWorkerQueue, generateSessionNameActivity, {
teamId,
agentId,
prompt,
sessionId,
context,
}).catch((err) => {
console.error('Failed to run generateSessionNameActivity:', err)
})
}
// 7. Update Placeholder Comment
if (placeholderCommentId) {
if (aiResult.text.trim() === '__NO_REPLY__') {
@@ -175,10 +246,25 @@ export async function agentChat(task: WorkflowTask): Promise<void> {
// Update placeholder comment with error message
if (placeholderCommentId && agentWorkerQueue) {
try {
const errorMessage =
err instanceof Error && err.message.startsWith('AI error:')
? `AI error: ${err.message.substring(9)}`
: 'Sorry, I encountered an error while processing your request.'
let errorMessage = 'Sorry, I encountered an error while processing your request.'
if (err instanceof Error) {
if (err.message.startsWith('AI error:')) {
errorMessage = `AI error: ${err.message.substring(9)}`
} else if (
err.message.includes('aborted') ||
err.message.includes('abort') ||
err.name === 'AbortError' ||
err.message.includes('cancel')
) {
errorMessage = 'Agent execution stopped.'
}
} else if (
String(err).includes('aborted') ||
String(err).includes('abort') ||
String(err).includes('cancel')
) {
errorMessage = 'Agent execution stopped.'
}
await executeActivity(agentWorkerQueue, updateCommentActivity, {
commentId: placeholderCommentId,
+1 -1
View File
@@ -1,6 +1,6 @@
{
"name": "@shumai/api",
"version": "0.1.6",
"version": "0.2.2",
"private": true,
"type": "module",
"main": "./src/index.ts",
+14 -2
View File
@@ -27,6 +27,7 @@ describe('Agent API', () => {
model: 'model1',
thinkingLevel: 'high',
systemPrompt: 'prompt',
deniedTools: ['bash'],
},
skills: [],
},
@@ -43,6 +44,7 @@ describe('Agent API', () => {
expect(data).toHaveLength(1)
expect(data[0].name).toBe('Bot 1')
expect(data[0].providerId).toBe('prov1')
expect(data[0].deniedTools).toEqual(['bash'])
})
})
@@ -55,7 +57,13 @@ describe('Agent API', () => {
providerId: 'prov1',
modelId: 'model1',
soul: 'New Soul',
config: { provider: 'prov1', model: 'model1', thinkingLevel: 'low', systemPrompt: 'p' },
config: {
provider: 'prov1',
model: 'model1',
thinkingLevel: 'low',
systemPrompt: 'p',
deniedTools: ['bash'],
},
skills: [],
}
// eslint-disable-next-line @typescript-eslint/no-explicit-any
@@ -73,6 +81,7 @@ describe('Agent API', () => {
systemPrompt: 'p',
soul: 'New Soul',
skills: ['skill1'],
deniedTools: ['bash'],
}),
})
@@ -80,6 +89,7 @@ describe('Agent API', () => {
const data = await res.json()
expect(data.name).toBe('New Agent')
expect(data.soul).toBe('New Soul')
expect(data.deniedTools).toEqual(['bash'])
})
it('validates request', async () => {
@@ -105,7 +115,7 @@ describe('Agent API', () => {
providerId: 'prov1',
modelId: 'model1',
soul: 'Updated Soul',
config: { provider: 'prov1', model: 'model1' },
config: { provider: 'prov1', model: 'model1', deniedTools: ['bash'] },
skills: [],
}
// eslint-disable-next-line @typescript-eslint/no-explicit-any
@@ -121,6 +131,7 @@ describe('Agent API', () => {
modelId: 'model1',
soul: 'Updated Soul',
skills: [],
deniedTools: ['bash'],
}),
})
@@ -128,6 +139,7 @@ describe('Agent API', () => {
const data = await res.json()
expect(data.name).toBe('Updated Agent')
expect(data.soul).toBe('Updated Soul')
expect(data.deniedTools).toEqual(['bash'])
})
})
+3
View File
@@ -47,6 +47,7 @@ const route = new Hono<{ Variables: { user: User } }>()
skillId: s.skillId,
skill: s.skill,
})),
deniedTools: config.deniedTools || [],
}
}),
)
@@ -89,6 +90,7 @@ const route = new Hono<{ Variables: { user: User } }>()
skillId: s.skillId,
skill: s.skill,
})),
deniedTools: config.deniedTools || [],
}
return c.json(info, 200)
@@ -127,6 +129,7 @@ const route = new Hono<{ Variables: { user: User } }>()
skillId: s.skillId,
skill: s.skill,
})),
deniedTools: config.deniedTools || [],
}
return c.json(info, 200)
+214
View File
@@ -0,0 +1,214 @@
import { describe, expect, it, vi, beforeEach } from 'vitest'
import { Hono } from 'hono'
import chatRoute from './chat'
import { chatService } from '@shumai/core/src/chat/chat'
import { authzService } from '@shumai/core/src/authz/authz'
import type { ChatMessage } from '@shumai/dtos'
vi.mock('@shumai/core/src/chat/chat', async (importOriginal) => {
const original = await importOriginal<typeof import('@shumai/core/src/chat/chat')>()
return {
...original,
chatService: {
listSessions: vi.fn(),
listMessages: vi.fn(),
deleteSession: vi.fn(),
startOrContinueChat: vi.fn(),
getNewSessionMessages: vi.fn(),
getChatWorkflowStatus: vi.fn(),
abortSession: vi.fn(),
},
}
})
vi.mock('@shumai/core/src/authz/authz')
describe('Chat API', () => {
const app = new Hono<{ Variables: { user: { id: string; name: string } } }>()
.use('*', async (c, next) => {
c.set('user', { id: 'user1', name: 'Test User' })
await next()
})
.route('/', chatRoute)
beforeEach(() => {
vi.restoreAllMocks()
vi.mocked(authzService.hasPermission).mockResolvedValue(undefined)
})
describe('GET /teams/:teamId/chat/sessions', () => {
it('returns list of sessions', async () => {
const mockSessions = {
data: [
{
id: 'session1',
agentId: 'default',
userId: 'user1',
assetId: 'asset1',
userCommentId: 'comment1',
createdAt: '2026-07-07T10:00:00Z',
updatedAt: '2026-07-07T10:00:00Z',
},
],
pageInfo: { total: 1 },
}
// eslint-disable-next-line @typescript-eslint/no-explicit-any
vi.mocked(chatService.listSessions).mockResolvedValue(mockSessions as any)
const res = await app.request('/teams/team1/chat/sessions?first=10')
expect(res.status).toBe(200)
const data = await res.json()
expect(data.data).toHaveLength(1)
expect(data.data[0].id).toBe('session1')
expect(chatService.listSessions).toHaveBeenCalledWith('user1', 'team1', expect.any(Object))
})
})
describe('GET /teams/:teamId/chat/sessions/:sessionId/messages', () => {
it('returns list of messages', async () => {
const mockMessages = [
{
id: 'msg1',
role: 'user',
content: 'hello',
timestamp: '2026-07-07T10:00:00Z',
entry: {},
},
]
// eslint-disable-next-line @typescript-eslint/no-explicit-any
vi.mocked(chatService.listMessages).mockResolvedValue(mockMessages as any)
const res = await app.request('/teams/team1/chat/sessions/session1/messages')
expect(res.status).toBe(200)
const data = await res.json()
expect(data).toHaveLength(1)
expect(data[0].content).toBe('hello')
expect(chatService.listMessages).toHaveBeenCalledWith('user1', 'team1', 'session1')
})
})
describe('DELETE /teams/:teamId/chat/sessions/:sessionId', () => {
it('deletes a session', async () => {
vi.mocked(chatService.deleteSession).mockResolvedValue(undefined)
const res = await app.request('/teams/team1/chat/sessions/session1', {
method: 'DELETE',
})
expect(res.status).toBe(204)
expect(chatService.deleteSession).toHaveBeenCalledWith('user1', 'team1', 'session1')
})
})
describe('POST /teams/:teamId/chat', () => {
it('starts or continues chat and returns stream', async () => {
vi.mocked(chatService.startOrContinueChat).mockResolvedValue({
sessionId: 'session1',
taskId: 'task1',
})
vi.mocked(chatService.getNewSessionMessages)
.mockResolvedValueOnce({
messages: [
{
id: 'entry1',
role: 'user',
content: 'hi',
} as unknown as ChatMessage,
],
lastEntryId: 'entry1',
})
.mockResolvedValue({
messages: [],
lastEntryId: 'entry1',
})
vi.mocked(chatService.getChatWorkflowStatus).mockResolvedValue({
status: 'completed',
output: {},
})
const res = await app.request('/teams/team1/chat', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
agentId: 'agent1',
textPrompt: 'hi',
}),
})
expect(res.status).toBe(200)
expect(res.headers.get('content-type')).toContain('text/event-stream')
const text = await res.text()
expect(text).toContain('data:')
expect(text).toContain('entry1')
expect(chatService.startOrContinueChat).toHaveBeenCalledWith(
expect.objectContaining({ id: 'user1' }),
'team1',
expect.any(Object),
)
})
it('sends ping keep-alive SSE events when there are no new messages', async () => {
vi.mocked(chatService.startOrContinueChat).mockResolvedValue({
sessionId: 'session1',
taskId: 'task1',
})
vi.mocked(chatService.getNewSessionMessages).mockResolvedValue({
messages: [],
lastEntryId: null,
})
vi.mocked(chatService.getChatWorkflowStatus).mockResolvedValue({
status: 'completed',
output: {},
})
const res = await app.request('/teams/team1/chat', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
agentId: 'agent1',
textPrompt: 'hi',
}),
})
expect(res.status).toBe(200)
expect(res.headers.get('content-type')).toContain('text/event-stream')
const text = await res.text()
expect(text).toContain('data:')
expect(text).toContain('{"type":"ping"}')
})
})
describe('POST /teams/:teamId/chat/sessions/:sessionId/abort', () => {
it('aborts an active chat session and updates task status', async () => {
vi.mocked(chatService.abortSession).mockResolvedValue(undefined)
const res = await app.request('/teams/team1/chat/sessions/session1/abort', {
method: 'POST',
})
expect(res.status).toBe(200)
const data = await res.json()
expect(data).toEqual({ success: true })
expect(chatService.abortSession).toHaveBeenCalledWith('user1', 'session1')
})
it('returns 404 if no active task is found for the session', async () => {
vi.mocked(chatService.abortSession).mockRejectedValue(
new Error('No active execution found for this session'),
)
const res = await app.request('/teams/team1/chat/sessions/session1/abort', {
method: 'POST',
})
expect(res.status).toBe(404)
const data = await res.json()
expect(data.error).toBe('No active execution found for this session')
})
})
})
+193
View File
@@ -0,0 +1,193 @@
import { Hono } from 'hono'
import { zValidator } from '@hono/zod-validator'
import { streamSSE } from 'hono/streaming'
import { chatService } from '@shumai/core/src/chat/chat'
import { authzService, Permission, ResourceType } from '@shumai/core/src/authz/authz'
import { chatRequestSchema, paginationParamsSchema } from '@shumai/dtos'
import type { Prisma } from '@shumai/db'
type User = Prisma.UserGetPayload<Record<string, never>>
const route = new Hono<{ Variables: { user: User } }>()
.get('/teams/:teamId/chat/sessions', zValidator('query', paginationParamsSchema), async (c) => {
const user = c.get('user')
const teamId = c.req.param('teamId')
const query = c.req.valid('query')
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Team,
id: teamId,
})
const res = await chatService.listSessions(user.id, teamId, query)
return c.json(res, 200)
})
.get('/teams/:teamId/chat/sessions/:sessionId/messages', async (c) => {
const user = c.get('user')
const teamId = c.req.param('teamId')
const sessionId = c.req.param('sessionId')
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Team,
id: teamId,
})
const res = await chatService.listMessages(user.id, teamId, sessionId)
return c.json(res, 200)
})
.delete('/teams/:teamId/chat/sessions/:sessionId', async (c) => {
const user = c.get('user')
const teamId = c.req.param('teamId')
const sessionId = c.req.param('sessionId')
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Team,
id: teamId,
})
await chatService.deleteSession(user.id, teamId, sessionId)
return new Response(null, { status: 204 })
})
.post('/teams/:teamId/chat', zValidator('json', chatRequestSchema), async (c) => {
const user = c.get('user')
const teamId = c.req.param('teamId')
const req = c.req.valid('json')
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Team,
id: teamId,
})
const { sessionId, taskId } = await chatService.startOrContinueChat(user, teamId, req)
c.header('x-session-id', sessionId)
return streamSSE(c, async (stream) => {
// Send session ID as the first event
await stream.writeSSE({
data: JSON.stringify({
type: 'session',
sessionId,
}),
})
let lastEntryId = ''
const timeoutMs = 5 * 60 * 1000 // 5 minutes timeout
const startTime = Date.now()
while (Date.now() - startTime < timeoutMs) {
if (stream.aborted) {
break
}
// Query new entries from the DB via chatService
const { messages, lastEntryId: newLastId } = await chatService.getNewSessionMessages(
sessionId,
lastEntryId || undefined,
)
if (messages.length > 0) {
for (const message of messages) {
await stream.writeSSE({
data: JSON.stringify({
type: 'entry',
entry: message,
}),
})
}
} else {
await stream.writeSSE({
data: JSON.stringify({
type: 'ping',
}),
})
}
if (newLastId) {
lastEntryId = newLastId
}
// Check if workflow has finished via chatService
const task = await chatService.getChatWorkflowStatus(taskId)
if (task && (task.status === 'completed' || task.status === 'failed')) {
// Fetch any remaining entries one last time
const { messages: finalMessages } = await chatService.getNewSessionMessages(
sessionId,
lastEntryId || undefined,
)
for (const message of finalMessages) {
await stream.writeSSE({
data: JSON.stringify({
type: 'entry',
entry: message,
}),
})
}
// Send done status
const output = task.output
await stream.writeSSE({
data: JSON.stringify({
type: 'done',
status: task.status,
error:
task.status === 'failed'
? (output?.error as string) || 'Workflow failed'
: undefined,
}),
})
break
}
// Wait before polling again
await stream.sleep(1000)
}
if (Date.now() - startTime >= timeoutMs) {
await stream.writeSSE({
data: JSON.stringify({
type: 'done',
status: 'failed',
error: 'Request timed out',
}),
})
}
})
})
.post('/teams/:teamId/chat/sessions/:sessionId/abort', async (c) => {
const user = c.get('user')
const teamId = c.req.param('teamId')
const sessionId = c.req.param('sessionId')
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Team,
id: teamId,
})
try {
await chatService.abortSession(user.id, sessionId)
return c.json({ success: true }, 200)
} catch (error: unknown) {
const err = error as Error
if (
err.message === 'Session not found' ||
err.message === 'Unauthorized session access' ||
err.message === 'No active execution found for this session'
) {
return c.json({ error: err.message }, 404)
}
throw error
}
})
export default route
+75 -3
View File
@@ -56,6 +56,8 @@ describe('file api', () => {
vi.spyOn(assetService, 'updateAssetName').mockImplementation(vi.fn())
vi.spyOn(assetService, 'deleteAssets').mockImplementation(vi.fn())
vi.spyOn(assetService, 'createComment').mockImplementation(vi.fn())
vi.spyOn(assetService, 'completeComment').mockImplementation(vi.fn())
vi.spyOn(assetService, 'deleteComment').mockImplementation(vi.fn())
vi.spyOn(assetService, 'listComments').mockImplementation(vi.fn())
vi.spyOn(assetService, 'restoreAssets').mockImplementation(vi.fn())
vi.spyOn(metadataService, 'updateAssetMetadata').mockImplementation(vi.fn())
@@ -77,7 +79,7 @@ describe('file api', () => {
fileCount: 1,
type: 'file',
status: 'active',
mediaType: 'image/jpeg',
proxyType: 'image',
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
})
@@ -105,7 +107,7 @@ describe('file api', () => {
fileCount: 1,
type: 'file',
status: 'active',
mediaType: 'image/jpeg',
proxyType: 'image',
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
})
@@ -185,6 +187,8 @@ describe('file api', () => {
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
sessionId: null,
isCompleted: false,
completionLastChangedBy: null,
})
// Mock assetService.getAsset to return teamId for notification
@@ -234,6 +238,8 @@ describe('file api', () => {
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
sessionId: null,
isCompleted: false,
completionLastChangedBy: null,
})
// Mock assetService.getAsset to return teamId for notification
@@ -290,6 +296,8 @@ describe('file api', () => {
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
sessionId: null,
isCompleted: false,
completionLastChangedBy: null,
},
],
pageInfo: { cursor: 'cursor', total: 1 },
@@ -311,6 +319,70 @@ describe('file api', () => {
})
})
it('POST /comments/:commentId/complete', async () => {
vi.mocked(assetService.completeComment).mockResolvedValue({
id: 'comment-id',
assetId: 'file-id',
message: 'hello',
annotations: null,
second: null,
creator: { id: 'user-id', name: 'Test User' },
replies: [],
attachments: [],
mentions: [],
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
sessionId: null,
isCompleted: true,
completionLastChangedBy: { id: 'user1', name: 'Test User' },
})
const app = new Hono().use('*', authMiddleware).route('/', fileRoute)
const res = await app.request('/comments/comment-id/complete', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({ isCompleted: true }),
})
expect(res.status).toBe(200)
const json = await res.json()
expect(json.isCompleted).toBe(true)
expect(json.completionLastChangedBy.id).toBe('user1')
expect(authzService.hasPermission).toHaveBeenCalledWith({
user: { id: 'user1', name: 'Test User' },
permission: Permission.Read,
type: ResourceType.Comment,
id: 'comment-id',
})
})
it('DELETE /comments/:commentId', async () => {
vi.mocked(assetService.deleteComment).mockResolvedValue(undefined)
const app = new Hono().use('*', authMiddleware).route('/', fileRoute)
const res = await app.request('/comments/comment-id', {
method: 'DELETE',
})
expect(res.status).toBe(200)
const json = await res.json()
expect(json.success).toBe(true)
expect(authzService.hasPermission).toHaveBeenCalledWith({
user: { id: 'user1', name: 'Test User' },
permission: Permission.Read,
type: ResourceType.Comment,
id: 'comment-id',
})
expect(assetService.deleteComment).toHaveBeenCalledWith({
commentId: 'comment-id',
userId: 'user1',
})
})
it('POST /teams/:teamId/files - image with compression', async () => {
vi.mocked(s3Service.putObject).mockResolvedValue(undefined)
@@ -428,7 +500,7 @@ describe('file api', () => {
fileCount: 1,
type: 'file',
status: 'active',
mediaType: 'image/jpeg',
proxyType: 'image',
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
})
+34
View File
@@ -16,6 +16,7 @@ import {
updateFileRequestSchema,
uploadFileRequestSchema,
getDownloadLinksRequestSchema,
completeCommentRequestSchema,
} from '@shumai/dtos'
import { transcodeService } from '@shumai/core'
import fs from 'fs'
@@ -200,6 +201,39 @@ const route = new Hono<{ Variables: { user: User } }>()
const comments = await assetService.listComments(fileId, req)
return c.json(comments)
})
.post(
'/comments/:commentId/complete',
zValidator('json', completeCommentRequestSchema),
async (c) => {
const commentId = c.req.param('commentId')
const user = c.get('user')
const req = c.req.valid('json')
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Comment,
id: commentId,
})
const updated = await assetService.completeComment(commentId, req.isCompleted, user.id)
return c.json(updated)
},
)
.delete('/comments/:commentId', async (c) => {
const commentId = c.req.param('commentId')
const user = c.get('user')
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Comment,
id: commentId,
})
await assetService.deleteComment({ commentId, userId: user.id })
return c.json({ success: true })
})
.post('/files/restore', zValidator('json', restoreFilesRequestSchema), async (c) => {
const user = c.get('user')
const req = c.req.valid('json')
-6
View File
@@ -53,7 +53,6 @@ describe('folder api', () => {
fileCount: 1,
type: 'folder',
status: 'active',
mediaType: null,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
})
@@ -81,7 +80,6 @@ describe('folder api', () => {
fileCount: 1,
type: 'folder',
status: 'active',
mediaType: null,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
})
@@ -113,7 +111,6 @@ describe('folder api', () => {
fileCount: 0,
type: 'folder',
status: 'active',
mediaType: null,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
})
@@ -147,7 +144,6 @@ describe('folder api', () => {
fileCount: 0,
type: 'folder',
status: 'active',
mediaType: null,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
},
@@ -223,7 +219,6 @@ describe('folder api', () => {
fileCount: 0,
type: 'folder',
status: 'active',
mediaType: null,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
},
@@ -261,7 +256,6 @@ describe('folder api', () => {
fileCount: 1,
type: 'folder',
status: 'active',
mediaType: null,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
})
+2
View File
@@ -1,6 +1,7 @@
import type { Prisma } from '@shumai/db'
import { Hono } from 'hono'
import agentRoute from './agent'
import chatRoute from './chat'
import attachmentRoute from './attachment'
import authnRoute from './authn'
import collectionRoute from './collection'
@@ -54,6 +55,7 @@ const apiRoute = new Hono<{ Variables: { user: User } }>()
.route('/', fileRoute)
.route('/', folderRoute)
.route('/', agentRoute)
.route('/', chatRoute)
.route('/', attachmentRoute)
.route('/', uploadRoute)
.route('/', shareRoute)
+38
View File
@@ -47,6 +47,7 @@ describe('Skill API', () => {
config: null,
assetId: 'asset1',
hash: 'hash1',
permission: 'reviewer' as const,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
},
@@ -79,6 +80,7 @@ describe('Skill API', () => {
config: null,
assetId: 'asset1',
hash: 'hash1',
permission: 'reviewer' as const,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
}
@@ -110,6 +112,7 @@ describe('Skill API', () => {
config: null,
assetId: 'asset1',
hash: 'hash1',
permission: 'reviewer' as const,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
}
@@ -174,6 +177,7 @@ describe('Skill API', () => {
config: mockConfig,
assetId: 'asset1',
hash: 'hash1',
permission: 'reviewer' as const,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
}
@@ -197,4 +201,38 @@ describe('Skill API', () => {
expect(skillService.updateSkillConfig).toHaveBeenCalledWith('skill1', mockConfig)
})
})
describe('PATCH /skills/:id/permission', () => {
test('Success', async () => {
const mockSkill = {
id: 'skill1',
name: 'Skill 1',
description: 'Desc 1',
config: null,
assetId: 'asset1',
hash: 'hash1',
permission: 'owner' as const,
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
}
vi.spyOn(skillService, 'updateSkillPermission').mockResolvedValue(mockSkill)
const res = await app.request('/skills/skill1/permission', {
method: 'PATCH',
headers: { 'Content-Type': 'application/json', Authorization: 'Bearer test' },
body: JSON.stringify({ permission: 'owner' }),
})
expect(res.status).toBe(200)
expect(await res.json()).toEqual(mockSkill)
expect(authzService.hasPermission).toHaveBeenCalledWith(
expect.objectContaining({
type: ResourceType.Skill,
id: 'skill1',
permission: Permission.Admin,
}),
)
expect(skillService.updateSkillPermission).toHaveBeenCalledWith('skill1', 'owner')
})
})
})
+24 -1
View File
@@ -2,7 +2,11 @@ import { Hono } from 'hono'
import { zValidator } from '@hono/zod-validator'
import { authzService, Permission, ResourceType } from '@shumai/core/src/authz/authz'
import { skillService } from '@shumai/core/src/skill/skill'
import { upsertSkillRequestSchema, updateSkillConfigRequestSchema } from '@shumai/dtos'
import {
upsertSkillRequestSchema,
updateSkillConfigRequestSchema,
updateSkillPermissionRequestSchema,
} from '@shumai/dtos'
import type { Prisma } from '@shumai/db'
type User = Prisma.UserGetPayload<Record<string, never>>
@@ -73,5 +77,24 @@ const route = new Hono<{ Variables: { user: User } }>()
const skill = await skillService.updateSkillConfig(id, req.config)
return c.json(skill)
})
.patch(
'/skills/:id/permission',
zValidator('json', updateSkillPermissionRequestSchema),
async (c) => {
const user = c.get('user')
const id = c.req.param('id')
const req = c.req.valid('json')
await authzService.hasPermission({
user,
permission: Permission.Admin,
type: ResourceType.Skill,
id,
})
const skill = await skillService.updateSkillPermission(id, req.permission)
return c.json(skill)
},
)
export default route
+68
View File
@@ -37,6 +37,13 @@ vi.mock('@shumai/core/src/user/api-token', () => ({
},
}))
vi.mock('@shumai/core/src/ai-usage/ai-usage', () => ({
aiUsageService: {
recordUsage: vi.fn(),
getTeamUsageStats: vi.fn(),
},
}))
describe('team api', () => {
const app = new Hono().use('*', authMiddleware).route('/', teamRoute)
let mockCreateTeam: any // eslint-disable-line @typescript-eslint/no-explicit-any
@@ -403,5 +410,66 @@ describe('team api', () => {
expect(data).toEqual({ success: true })
expect(apiTokenService.deleteToken).toHaveBeenCalledWith('user1', 'token1')
})
it('GET /teams/:teamId/ai-usage returns team usage stats for owner', async () => {
const { aiUsageService } = await import('@shumai/core/src/ai-usage/ai-usage')
vi.mocked(aiUsageService.getTeamUsageStats).mockResolvedValue({
timeframe: '30d',
team: {
inputTokens: 1000,
outputTokens: 500,
cacheReadTokens: 200,
totalTokens: 1500,
cost: 0.005,
},
})
const res = await app.request('/teams/t1/ai-usage?timeframe=30d')
expect(res.status).toBe(200)
const data = await res.json()
expect(data.team.totalTokens).toBe(1500)
expect(authzService.hasPermission).toHaveBeenCalledWith(
expect.objectContaining({
permission: Permission.Admin,
type: ResourceType.Team,
id: 't1',
}),
)
expect(aiUsageService.getTeamUsageStats).toHaveBeenCalledWith({
teamId: 't1',
timeframe: '30d',
userId: undefined,
})
})
it('GET /teams/:teamId/ai-usage?userId=user1 returns specific member usage stats', async () => {
const { aiUsageService } = await import('@shumai/core/src/ai-usage/ai-usage')
vi.mocked(aiUsageService.getTeamUsageStats).mockResolvedValue({
timeframe: '7d',
member: {
userId: 'user1',
userName: 'Test User',
userEmail: 'user1@example.com',
role: 'owner',
inputTokens: 400,
outputTokens: 200,
cacheReadTokens: 50,
totalTokens: 600,
cost: 0.002,
},
})
const res = await app.request('/teams/t1/ai-usage?timeframe=7d&userId=user1')
expect(res.status).toBe(200)
const data = await res.json()
expect(data.member.userId).toBe('user1')
expect(aiUsageService.getTeamUsageStats).toHaveBeenCalledWith({
teamId: 't1',
timeframe: '7d',
userId: 'user1',
})
})
})
})
+22 -1
View File
@@ -15,8 +15,9 @@ import {
updateTeamMemberRoleRequestSchema,
createApiTokenRequestSchema,
} from '@shumai/dtos'
import { updateUserMetadataRequestSchema } from '@shumai/dtos'
import { updateUserMetadataRequestSchema, getTeamAiUsageQuerySchema } from '@shumai/dtos'
import { apiTokenService } from '@shumai/core/src/user/api-token'
import { aiUsageService } from '@shumai/core/src/ai-usage/ai-usage'
import type { Prisma } from '@shumai/db'
type User = Prisma.UserGetPayload<Record<string, never>>
@@ -313,5 +314,25 @@ const route = new Hono<{ Variables: { user: User } }>()
await apiTokenService.deleteToken(user.id, tokenId)
return c.json({ success: true })
})
.get('/teams/:teamId/ai-usage', zValidator('query', getTeamAiUsageQuerySchema), async (c) => {
const user = c.get('user')
const teamId = c.req.param('teamId')
const { timeframe, userId } = c.req.valid('query')
await authzService.hasPermission({
user,
permission: Permission.Admin,
type: ResourceType.Team,
id: teamId,
})
const stats = await aiUsageService.getTeamUsageStats({
teamId,
timeframe,
userId,
})
return c.json(stats)
})
export default route
+6 -2
View File
@@ -1,22 +1,25 @@
{
"name": "@shumai/core",
"version": "0.1.6",
"version": "0.2.2",
"private": true,
"type": "module",
"main": "./src/index.ts",
"types": "./src/index.ts",
"dependencies": {
"@earendil-works/pi-agent-core": "0.78.0",
"@shumai/db": "workspace:*",
"@shumai/dtos": "workspace:*",
"@sinclair/typebox": "0.34.49",
"adm-zip": "0.5.17",
"better-auth": "1.6.11",
"csv-parse": "7.0.1",
"date-fns": "4.4.0",
"dotenv": "16.6.1",
"dotenv-expand": "13.0.0",
"extract-zip": "2.0.1",
"hono": "4.12.23",
"jittered-fractional-indexing": "1.0.1",
"pdfkit": "0.19.1",
"pino": "10.3.1",
"sharp": "0.34.5",
"tweetnacl": "1.0.3",
@@ -26,6 +29,7 @@
},
"devDependencies": {
"@types/adm-zip": "0.5.8",
"@types/extract-zip": "2.0.3"
"@types/extract-zip": "2.0.3",
"@types/pdfkit": "0.17.6"
}
}
+24
View File
@@ -126,6 +126,27 @@ describe('AgentService', () => {
expect(agent?.skills?.[0].skillId).toBe(skill.id)
})
test('Create Agent with Denied Tools', async () => {
const db = prisma
const svc = new AgentService()
const { team, provider, model } = await setupTestData(db)
const agent = await svc.createAgent({
teamId: team.id,
name: 'Denied Tools Agent',
type: 'chat',
enabled: true,
providerId: provider.id,
modelId: model.id,
thinkingLevel: 'medium',
deniedTools: ['bash', 'screenshot'],
})
expect(agent?.id).toBeDefined()
const config = agent?.config as unknown as PrismaJson.AgentConfig
expect(config.deniedTools).toEqual(['bash', 'screenshot'])
})
test('Create Agent validation fails if model mismatch', async () => {
const db = prisma
const svc = new AgentService()
@@ -186,8 +207,11 @@ describe('AgentService', () => {
thinkingLevel: 'high',
soul: 'New soul',
skills: [skill.id], // Re-enable one
deniedTools: ['list_assets'],
})
expect(reUpdated.skills?.length).toBe(1)
const config = reUpdated.config as unknown as PrismaJson.AgentConfig
expect(config.deniedTools).toEqual(['list_assets'])
})
test('Create Agent with presigned URL extracts S3 key', async () => {
+4
View File
@@ -114,6 +114,7 @@ export class AgentService {
systemPrompt,
soul,
skills,
deniedTools,
} = params
if (type === 'embedding' || type === 'autofill') {
@@ -152,6 +153,7 @@ export class AgentService {
model: modelId || '',
thinkingLevel,
systemPrompt,
deniedTools: deniedTools || [],
}
const user = await tx.user.create({
@@ -211,6 +213,7 @@ export class AgentService {
systemPrompt,
soul,
skills,
deniedTools,
} = params
const agent = await this.prismaClient.agent.findUnique({
@@ -251,6 +254,7 @@ export class AgentService {
model: modelId || '',
thinkingLevel,
systemPrompt,
deniedTools: deniedTools || [],
}
return this.prismaClient.$transaction(async (tx) => {
+313
View File
@@ -0,0 +1,313 @@
import { describe, expect, it, vi } from 'vitest'
import { prisma } from '@shumai/db'
import { setupTestDbHooks } from '@shumai/db/test'
import { aiUsageService } from './ai-usage'
import { HTTPException } from 'hono/http-exception'
describe('AiUsageService', () => {
setupTestDbHooks()
async function createTestTeamAndUser() {
const user = await prisma.user.create({
data: {
name: 'Usage Test User',
email: `usage-test-${Date.now()}-${Math.random()}@example.com`,
},
})
const team = await prisma.team.create({
data: {
name: 'Usage Test Team',
},
})
const teamMember = await prisma.teamMember.create({
data: {
teamId: team.id,
userId: user.id,
role: 'owner',
},
})
return { user, team, teamMember }
}
it('records usage for both team total and specific member', async () => {
const { team, user } = await createTestTeamAndUser()
const now = new Date()
await aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 100,
outputTokens: 50,
cacheReadTokens: 20,
totalTokens: 150,
cost: 0.001,
timestamp: now,
})
// Verify database rows
const records = await prisma.aiUsage.findMany({
where: { teamId: team.id },
})
expect(records).toHaveLength(2)
const teamRecord = records.find((r) => r.userId === null)
expect(teamRecord).toBeDefined()
expect(teamRecord?.inputTokens).toBe(100)
expect(teamRecord?.outputTokens).toBe(50)
expect(teamRecord?.cacheReadTokens).toBe(20)
expect(teamRecord?.totalTokens).toBe(150)
expect(teamRecord?.cost).toBeCloseTo(0.001)
const memberRecord = records.find((r) => r.userId === user.id)
expect(memberRecord).toBeDefined()
expect(memberRecord?.inputTokens).toBe(100)
expect(memberRecord?.outputTokens).toBe(50)
expect(memberRecord?.cacheReadTokens).toBe(20)
expect(memberRecord?.totalTokens).toBe(150)
expect(memberRecord?.cost).toBeCloseTo(0.001)
})
it('increments existing hourly buckets for team and member', async () => {
const { team, user } = await createTestTeamAndUser()
const now = new Date()
await aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 100,
outputTokens: 50,
cacheReadTokens: 10,
totalTokens: 150,
cost: 0.001,
timestamp: now,
})
await aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 200,
outputTokens: 100,
cacheReadTokens: 30,
totalTokens: 300,
cost: 0.002,
timestamp: now,
})
const teamStats = await aiUsageService.getTeamUsageStats({
teamId: team.id,
timeframe: '1h',
})
expect(teamStats.team).toBeDefined()
expect(teamStats.team?.inputTokens).toBe(300)
expect(teamStats.team?.outputTokens).toBe(150)
expect(teamStats.team?.cacheReadTokens).toBe(40)
expect(teamStats.team?.totalTokens).toBe(450)
expect(teamStats.team?.cost).toBeCloseTo(0.003)
const memberStats = await aiUsageService.getTeamUsageStats({
teamId: team.id,
timeframe: '1h',
userId: user.id,
})
expect(memberStats.member).toBeDefined()
expect(memberStats.member?.userId).toBe(user.id)
expect(memberStats.member?.inputTokens).toBe(300)
expect(memberStats.member?.outputTokens).toBe(150)
expect(memberStats.member?.cacheReadTokens).toBe(40)
expect(memberStats.member?.totalTokens).toBe(450)
expect(memberStats.member?.cost).toBeCloseTo(0.003)
})
it('handles background system tasks with no userId', async () => {
const { team } = await createTestTeamAndUser()
await aiUsageService.recordUsage({
teamId: team.id,
userId: null,
inputTokens: 50,
outputTokens: 25,
cacheReadTokens: 5,
totalTokens: 75,
cost: 0.0005,
})
const records = await prisma.aiUsage.findMany({
where: { teamId: team.id },
})
expect(records).toHaveLength(1)
expect(records[0].userId).toBeNull()
expect(records[0].inputTokens).toBe(50)
})
it('filters statistics correctly by timeframe', async () => {
vi.useFakeTimers()
const now = new Date('2026-07-23T15:30:00Z')
vi.setSystemTime(now)
try {
const { team, user } = await createTestTeamAndUser()
// 5 minutes ago (within current 1h bucket)
const recent = new Date(now.getTime() - 5 * 60 * 1000)
// 2 hours ago (within 24h, outside 1h)
const twoHoursAgo = new Date(now.getTime() - 2 * 60 * 60 * 1000)
// 3 days ago (within 7d, outside 24h)
const threeDaysAgo = new Date(now.getTime() - 3 * 24 * 60 * 60 * 1000)
await aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 100,
outputTokens: 50,
cacheReadTokens: 10,
totalTokens: 150,
cost: 0.001,
timestamp: recent,
})
await aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 200,
outputTokens: 100,
cacheReadTokens: 20,
totalTokens: 300,
cost: 0.002,
timestamp: twoHoursAgo,
})
await aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 400,
outputTokens: 200,
cacheReadTokens: 40,
totalTokens: 600,
cost: 0.004,
timestamp: threeDaysAgo,
})
// 1h timeframe should only include recent
const stats1h = await aiUsageService.getTeamUsageStats({
teamId: team.id,
timeframe: '1h',
})
expect(stats1h.team?.inputTokens).toBe(100)
// 24h timeframe should include recent + 2 hours ago
const stats24h = await aiUsageService.getTeamUsageStats({
teamId: team.id,
timeframe: '24h',
})
expect(stats24h.team?.inputTokens).toBe(300)
// 7d timeframe should include all 3
const stats7d = await aiUsageService.getTeamUsageStats({
teamId: team.id,
timeframe: '7d',
})
expect(stats7d.team?.inputTokens).toBe(700)
} finally {
vi.useRealTimers()
}
})
it('throws 404 when querying non-existent team member', async () => {
const { team } = await createTestTeamAndUser()
await expect(
aiUsageService.getTeamUsageStats({
teamId: team.id,
timeframe: '30d',
userId: 'non-existent-user-id',
}),
).rejects.toThrow(HTTPException)
})
it('targets only current hour bucket for 1h timeframe instead of including previous hour bucket', async () => {
const { team, user } = await createTestTeamAndUser()
// 10:15 AM (previous hour)
const prevHour = new Date('2026-07-23T10:15:00Z')
// 11:10 AM (current hour)
const currHour = new Date('2026-07-23T11:10:00Z')
await aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 100,
outputTokens: 0,
cacheReadTokens: 0,
totalTokens: 100,
cost: 0.001,
timestamp: prevHour,
})
await aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 50,
outputTokens: 0,
cacheReadTokens: 0,
totalTokens: 50,
cost: 0.0005,
timestamp: currHour,
})
vi.useFakeTimers()
vi.setSystemTime(new Date('2026-07-23T11:15:00Z'))
try {
const stats1h = await aiUsageService.getTeamUsageStats({
teamId: team.id,
timeframe: '1h',
})
expect(stats1h.team?.inputTokens).toBe(50)
} finally {
vi.useRealTimers()
}
})
it('handles concurrent recordUsage calls for the same team and user without duplication or race errors', async () => {
const { team, user } = await createTestTeamAndUser()
const now = new Date()
await Promise.all([
aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 100,
outputTokens: 50,
cacheReadTokens: 10,
totalTokens: 150,
cost: 0.001,
timestamp: now,
}),
aiUsageService.recordUsage({
teamId: team.id,
userId: user.id,
inputTokens: 200,
outputTokens: 100,
cacheReadTokens: 20,
totalTokens: 300,
cost: 0.002,
timestamp: now,
}),
])
const stats = await aiUsageService.getTeamUsageStats({
teamId: team.id,
timeframe: '1h',
})
expect(stats.team?.inputTokens).toBe(300)
expect(stats.team?.outputTokens).toBe(150)
})
})
+263
View File
@@ -0,0 +1,263 @@
import { prisma, Prisma } from '@shumai/db'
import type { TeamUsageStatsResponse, Timeframe } from '@shumai/dtos'
import { HTTPException } from 'hono/http-exception'
export interface RecordUsageParams {
teamId: string
userId?: string | null
inputTokens: number
outputTokens: number
cacheReadTokens: number
totalTokens: number
cost: number
timestamp?: Date
}
export interface GetTeamUsageStatsParams {
teamId: string
timeframe: Timeframe
userId?: string
}
export class AiUsageService {
private async incrementBucket(params: {
teamId: string
userId: string | null
periodStart: Date
inputTokens: number
outputTokens: number
cacheReadTokens: number
totalTokens: number
cost: number
}): Promise<void> {
if (params.userId !== null) {
await prisma.aiUsage.upsert({
where: {
// eslint-disable-next-line @typescript-eslint/naming-convention
teamId_userId_periodStart: {
teamId: params.teamId,
userId: params.userId,
periodStart: params.periodStart,
},
},
update: {
inputTokens: { increment: params.inputTokens },
outputTokens: { increment: params.outputTokens },
cacheReadTokens: { increment: params.cacheReadTokens },
totalTokens: { increment: params.totalTokens },
cost: { increment: params.cost },
},
create: {
teamId: params.teamId,
userId: params.userId,
periodStart: params.periodStart,
inputTokens: params.inputTokens,
outputTokens: params.outputTokens,
cacheReadTokens: params.cacheReadTokens,
totalTokens: params.totalTokens,
cost: params.cost,
},
})
} else {
const existing = await prisma.aiUsage.findFirst({
where: {
teamId: params.teamId,
userId: null,
periodStart: params.periodStart,
},
})
if (existing) {
await prisma.aiUsage.update({
where: { id: existing.id },
data: {
inputTokens: { increment: params.inputTokens },
outputTokens: { increment: params.outputTokens },
cacheReadTokens: { increment: params.cacheReadTokens },
totalTokens: { increment: params.totalTokens },
cost: { increment: params.cost },
},
})
} else {
try {
await prisma.aiUsage.create({
data: {
teamId: params.teamId,
userId: null,
periodStart: params.periodStart,
inputTokens: params.inputTokens,
outputTokens: params.outputTokens,
cacheReadTokens: params.cacheReadTokens,
totalTokens: params.totalTokens,
cost: params.cost,
},
})
} catch (err) {
if (err instanceof Prisma.PrismaClientKnownRequestError && err.code === 'P2002') {
await prisma.aiUsage.updateMany({
where: {
teamId: params.teamId,
userId: null,
periodStart: params.periodStart,
},
data: {
inputTokens: { increment: params.inputTokens },
outputTokens: { increment: params.outputTokens },
cacheReadTokens: { increment: params.cacheReadTokens },
totalTokens: { increment: params.totalTokens },
cost: { increment: params.cost },
},
})
} else {
throw err
}
}
}
}
}
async recordUsage(params: RecordUsageParams): Promise<void> {
const timestamp = params.timestamp ?? new Date()
const periodStart = new Date(timestamp)
periodStart.setMinutes(0, 0, 0)
// 1. Update/create Team Total record (userId = null)
await this.incrementBucket({
teamId: params.teamId,
userId: null,
periodStart,
inputTokens: params.inputTokens,
outputTokens: params.outputTokens,
cacheReadTokens: params.cacheReadTokens,
totalTokens: params.totalTokens,
cost: params.cost,
})
// 2. If triggered by a team member, update/create Member Usage record (userId = params.userId)
if (params.userId) {
await this.incrementBucket({
teamId: params.teamId,
userId: params.userId,
periodStart,
inputTokens: params.inputTokens,
outputTokens: params.outputTokens,
cacheReadTokens: params.cacheReadTokens,
totalTokens: params.totalTokens,
cost: params.cost,
})
}
}
async getTeamUsageStats(params: GetTeamUsageStatsParams): Promise<TeamUsageStatsResponse> {
const now = new Date()
const since = new Date(now)
switch (params.timeframe) {
case '1h':
// For 1h, target the current 1h bucket floored to start of current hour
since.setMinutes(0, 0, 0)
break
case '24h':
since.setDate(since.getDate() - 1)
since.setMinutes(0, 0, 0)
break
case '7d':
since.setDate(since.getDate() - 7)
since.setMinutes(0, 0, 0)
break
case '30d':
since.setDate(since.getDate() - 30)
since.setMinutes(0, 0, 0)
break
}
if (!params.userId) {
// Return Team-Level Usage (userId = null)
const teamRecords = await prisma.aiUsage.findMany({
where: {
teamId: params.teamId,
userId: null,
periodStart: { gte: since },
},
})
let inputTokens = 0
let outputTokens = 0
let cacheReadTokens = 0
let totalTokens = 0
let cost = 0
for (const r of teamRecords) {
inputTokens += r.inputTokens
outputTokens += r.outputTokens
cacheReadTokens += r.cacheReadTokens
totalTokens += r.totalTokens
cost += r.cost
}
return {
timeframe: params.timeframe,
team: {
inputTokens,
outputTokens,
cacheReadTokens,
totalTokens,
cost: Math.round((cost + Number.EPSILON) * 1000000) / 1000000,
},
}
} else {
// Return Specific Member Usage
const teamMember = await prisma.teamMember.findFirst({
where: {
teamId: params.teamId,
userId: params.userId,
},
include: { user: true },
})
if (!teamMember) {
throw new HTTPException(404, { message: 'Team member not found' })
}
const memberRecords = await prisma.aiUsage.findMany({
where: {
teamId: params.teamId,
userId: params.userId,
periodStart: { gte: since },
},
})
let inputTokens = 0
let outputTokens = 0
let cacheReadTokens = 0
let totalTokens = 0
let cost = 0
for (const r of memberRecords) {
inputTokens += r.inputTokens
outputTokens += r.outputTokens
cacheReadTokens += r.cacheReadTokens
totalTokens += r.totalTokens
cost += r.cost
}
return {
timeframe: params.timeframe,
member: {
userId: teamMember.user.id,
userName: teamMember.user.name,
userEmail: teamMember.user.email,
userImage: teamMember.user.image,
role: teamMember.role,
inputTokens,
outputTokens,
cacheReadTokens,
totalTokens,
cost: Math.round((cost + Number.EPSILON) * 1000000) / 1000000,
},
}
}
}
}
export const aiUsageService = new AiUsageService()
+186 -2
View File
@@ -1,6 +1,7 @@
import { describe, expect, it, vi, beforeEach, afterEach } from 'vitest'
import { prisma } from '@shumai/db'
import { setupTestDbHooks } from '@shumai/db/test'
import { HTTPException } from 'hono/http-exception'
import { AssetType, AssetStatus, Prisma } from '@shumai/db'
import { AssetService } from './asset'
@@ -721,6 +722,161 @@ describe('AssetService', () => {
expect(list.data[0].replies.length).toBe(1)
})
it('deleteComment validation and execution', async () => {
const { user, project, assets } = await setupBasicAssets()
// 1. Create a comment
const comment = await assetService.createComment({
assetId: assets.fileA1.id,
userId: user.id,
message: 'Self Comment',
attachmentIds: [],
})
// 2. Add user as a team member with owner role (team owner)
await prisma.teamMember.create({
data: {
teamId: project.teamId,
userId: user.id,
role: 'owner',
},
})
// Create another user
const otherUser = await prisma.user.create({
data: {
name: 'OtherUser',
type: 'human',
email: `other-${Date.now()}@example.com`,
},
})
// Add other user to the team as editor
await prisma.teamMember.create({
data: {
teamId: project.teamId,
userId: otherUser.id,
role: 'editor',
},
})
// Create comment by other user
const otherComment = await assetService.createComment({
assetId: assets.fileA1.id,
userId: otherUser.id,
message: 'Other Comment',
attachmentIds: [],
})
// 3. User (owner) should be able to delete their own comment
await expect(
assetService.deleteComment({ commentId: comment.id, userId: user.id }),
).resolves.not.toThrow()
// Verify it is deleted from DB
await expect(prisma.assetComment.findUnique({ where: { id: comment.id } })).resolves.toBeNull()
// 4. User (owner) should be able to delete other user's comment
await expect(
assetService.deleteComment({ commentId: otherComment.id, userId: user.id }),
).resolves.not.toThrow()
// Verify it is deleted from DB
await expect(
prisma.assetComment.findUnique({ where: { id: otherComment.id } }),
).resolves.toBeNull()
// Create a new comment by user
const comment2 = await assetService.createComment({
assetId: assets.fileA1.id,
userId: user.id,
message: 'Self Comment 2',
attachmentIds: [],
})
// 5. Other user (editor, non-creator) should NOT be able to delete user's comment
await expect(
assetService.deleteComment({ commentId: comment2.id, userId: otherUser.id }),
).rejects.toThrow(HTTPException)
// 6. Other user (editor, creator) should be able to delete their own comment
const otherComment2 = await assetService.createComment({
assetId: assets.fileA1.id,
userId: otherUser.id,
message: 'Other Comment 2',
attachmentIds: [],
})
await expect(
assetService.deleteComment({ commentId: otherComment2.id, userId: otherUser.id }),
).resolves.not.toThrow()
})
it('deleteComment cleans up S3 attachment files and database records', async () => {
const { user, project, assets } = await setupBasicAssets()
// 1. Add user as team owner
await prisma.teamMember.create({
data: {
teamId: project.teamId,
userId: user.id,
role: 'owner',
},
})
// 2. Create StorageKey and Asset for attachment
const storageKey = await prisma.storageKey.create({
data: {
key: 'attachments/test-attachment-file.png',
status: 'active',
},
})
const attachmentAsset = await prisma.asset.create({
data: {
name: 'test-attachment-file.png',
type: AssetType.file,
projectId: project.id,
parentId: assets.root.id,
creatorId: user.id,
storageKeyId: storageKey.id,
sizeByte: 100,
status: 'uploaded',
},
})
// 3. Create a comment with this attachment
const comment = await assetService.createComment({
assetId: assets.fileA1.id,
userId: user.id,
message: 'Comment with attachment',
attachmentIds: [attachmentAsset.id],
})
// Verify it is created in DB
const dbComment = await prisma.assetComment.findUnique({
where: { id: comment.id },
include: { attachments: true },
})
expect(dbComment?.attachments.length).toBe(1)
// 4. Delete the comment
await expect(
assetService.deleteComment({ commentId: comment.id, userId: user.id }),
).resolves.not.toThrow()
// 5. Verify the comment is deleted from DB
await expect(prisma.assetComment.findUnique({ where: { id: comment.id } })).resolves.toBeNull()
// 6. Verify that the attachment Asset and StorageKey are deleted from DB
await expect(prisma.asset.findUnique({ where: { id: attachmentAsset.id } })).resolves.toBeNull()
await expect(prisma.storageKey.findUnique({ where: { id: storageKey.id } })).resolves.toBeNull()
// 7. Verify S3 deletion was called
const { s3Service } = await import('@shumai/core/src/s3/s3')
expect(s3Service.deleteObject).toHaveBeenCalledWith(
'shumai',
'attachments/test-attachment-file.png',
)
})
it('can create and retrieve comments with video timestamp', async () => {
const { user, assets } = await setupBasicAssets()
@@ -855,6 +1011,31 @@ describe('AssetService', () => {
expect(listWithReplies.data[0].replies[1].id).toBe(r2.id)
})
it('can mark comments as complete and incomplete', async () => {
const { user, assets } = await setupBasicAssets()
const comment = await assetService.createComment({
assetId: assets.fileA1.id,
userId: user.id,
message: 'To be completed',
attachmentIds: [],
})
expect(comment.isCompleted).toBe(false)
expect(comment.completionLastChangedBy).toBeNull()
// Mark as complete
const completed = await assetService.completeComment(comment.id, true, user.id)
expect(completed.isCompleted).toBe(true)
expect(completed.completionLastChangedBy).toBeDefined()
expect(completed.completionLastChangedBy?.id).toBe(user.id)
// Mark as incomplete
const incompleted = await assetService.completeComment(comment.id, false, user.id)
expect(incompleted.isCompleted).toBe(false)
expect(incompleted.completionLastChangedBy?.id).toBe(user.id)
})
it('triggers AI workflow on bot mention', async () => {
const { user, project } = await setupBasicAssets()
@@ -1294,6 +1475,7 @@ describe('AssetService', () => {
mediaType: 'video/mp4',
status: 'processed',
media: {
proxyType: 'video',
poster: { key: 'poster-key' },
metadata: { duration: 262 },
// eslint-disable-next-line @typescript-eslint/no-explicit-any
@@ -1304,7 +1486,7 @@ describe('AssetService', () => {
const info = await assetService.getAsset({ assetId: video.id })
expect(info.preview).toBeDefined()
expect(info.preview?.mediaType).toBe('video/mp4')
expect(info.preview?.proxyType).toBe('video')
expect(info.preview?.duration).toBe(262)
})
@@ -1321,6 +1503,7 @@ describe('AssetService', () => {
mediaType: 'image/png',
status: 'processed',
media: {
proxyType: 'image',
thumbnail: { key: 'thumb-key' },
metadata: { originalWidth: 100, originalHeight: 80 },
// eslint-disable-next-line @typescript-eslint/no-explicit-any
@@ -1349,6 +1532,7 @@ describe('AssetService', () => {
mediaType: 'image/png',
status: 'processed',
media: {
proxyType: 'image',
thumbnail: { key: 'thumb-key' },
// eslint-disable-next-line @typescript-eslint/no-explicit-any
} as any,
@@ -1373,7 +1557,7 @@ describe('AssetService', () => {
expect(info.type).toBe(AssetType.symlink)
// Resolved from target
expect(info.sizeByte).toBe(5000)
expect(info.mediaType).toBe('image/png')
expect(info.proxyType).toBe('image')
expect(info.media).toBeDefined()
// eslint-disable-next-line @typescript-eslint/no-explicit-any
expect((info.media as any).thumbnail?.key).toBe('thumb-key')
+151 -5
View File
@@ -18,6 +18,7 @@ import {
AssetUserInfo,
} from '@shumai/dtos'
import { type Asset, AssetStatus, AssetType, Prisma, type StorageKey } from '@shumai/db'
import { HTTPException } from 'hono/http-exception'
import { logger } from '@shumai/core/src/logger'
import { PaginatedData, paginateQuery, PaginationParams } from '@shumai/core/src/pagination'
import { s3Service } from '@shumai/core/src/s3/s3'
@@ -48,10 +49,12 @@ type AssetWithIncludes = Prisma.AssetGetPayload<{
type CommentWithIncludes = Prisma.AssetCommentGetPayload<{
include: {
creator: true
completionLastChangedBy: true
attachments: { include: { asset: { include: { storageKey: true } } } }
replies: {
include: {
creator: true
completionLastChangedBy: true
attachments: { include: { asset: { include: { storageKey: true } } } }
}
}
@@ -99,6 +102,7 @@ export class AssetService {
metadataValues: true,
storageKey: true,
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -106,6 +110,7 @@ export class AssetService {
},
},
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -658,6 +663,7 @@ export class AssetService {
metadataValues: true,
storageKey: true,
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -665,6 +671,7 @@ export class AssetService {
},
},
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -742,6 +749,7 @@ export class AssetService {
metadataValues: true,
storageKey: true,
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -749,6 +757,7 @@ export class AssetService {
},
},
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -791,6 +800,7 @@ export class AssetService {
metadataValues: true,
storageKey: true,
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -798,6 +808,7 @@ export class AssetService {
},
},
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -933,6 +944,7 @@ export class AssetService {
metadataValues: true,
storageKey: true,
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -940,6 +952,7 @@ export class AssetService {
},
},
children: {
where: { isDeleted: false },
include: { creator: true, metadataValues: true, storageKey: true },
take: 3,
orderBy: { sortIndex: 'asc' },
@@ -1379,11 +1392,13 @@ export class AssetService {
where: { id: commentId },
include: {
creator: true,
completionLastChangedBy: true,
asset: { include: { storageKey: true } },
attachments: { include: { asset: { include: { storageKey: true } } } },
replies: {
include: {
creator: true,
completionLastChangedBy: true,
attachments: { include: { asset: { include: { storageKey: true } } } },
},
orderBy: { id: 'asc' },
@@ -1394,6 +1409,106 @@ export class AssetService {
return this.toCommentInfo(c)
}
async completeComment(
commentId: string,
isCompleted: boolean,
userId: string,
): Promise<CommentInfo> {
await this.prismaClient.assetComment.update({
where: { id: commentId },
data: {
isCompleted,
completionLastChangedById: userId,
},
})
return this.getComment(commentId)
}
async deleteComment({ commentId, userId }: { commentId: string; userId: string }): Promise<void> {
const comment = await this.prismaClient.assetComment.findUnique({
where: { id: commentId },
include: {
attachments: {
include: {
asset: {
include: {
storageKey: true,
},
},
},
},
asset: {
select: {
project: { select: { id: true, teamId: true } },
},
},
},
})
if (!comment) {
throw new HTTPException(404, { message: 'Comment not found' })
}
const teamId = comment.asset.project?.teamId
if (!teamId) {
throw new HTTPException(400, {
message: 'Data corruption: Comment asset is not associated with a project',
})
}
const member = await this.prismaClient.teamMember.findUnique({
where: {
teamIdUserId: {
teamId,
userId,
},
},
})
if (!member) {
throw new HTTPException(403, { message: 'User is not a member of the team' })
}
const isOwner = member.role === 'owner'
const isCreator = comment.creatorId === userId
if (!isOwner && !isCreator) {
throw new HTTPException(403, { message: 'You do not have permission to delete this comment' })
}
// Clean up attachment files from S3/storage and delete their DB records
const bucket = process.env.S3_BUCKET || 'shumai'
for (const att of comment.attachments) {
const asset = att.asset
if (asset.storageKey) {
try {
await s3Service.deleteObject(bucket, asset.storageKey.key)
} catch (err) {
logger.error({ err, assetId: asset.id }, 'Failed to delete attachment file from S3')
}
}
await this.prismaClient.asset.delete({
where: { id: asset.id },
})
if (asset.storageKey) {
await this.prismaClient.storageKey
.delete({
where: { id: asset.storageKey.id },
})
.catch(() => {
// Silently ignore if already deleted
})
}
}
await this.prismaClient.assetComment.delete({
where: { id: commentId },
})
}
async listComments(
assetId: string,
params: PaginationParams,
@@ -1406,11 +1521,13 @@ export class AssetService {
where: { assetId: resolvedAssetId, replyToId: null },
include: {
creator: true,
completionLastChangedBy: true,
asset: { include: { storageKey: true } },
attachments: { include: { asset: { include: { storageKey: true } } } },
replies: {
include: {
creator: true,
completionLastChangedBy: true,
attachments: { include: { asset: { include: { storageKey: true } } } },
},
orderBy: { id: 'asc' },
@@ -1534,6 +1651,7 @@ export class AssetService {
folderForPreview.children
) {
for (const child of folderForPreview.children) {
if (child.isDeleted) continue
let previewAsset = child as Asset
let mediaType = child.mediaType
@@ -1590,6 +1708,13 @@ export class AssetService {
'GET',
)
}
if (media && media.pdfTranscode?.key) {
media.pdfTranscode.url = await s3Service.presign(
process.env.S3_BUCKET || 'shumai',
media.pdfTranscode.key,
'GET',
)
}
const key = latestVersion.storageKey?.key
if (key) {
@@ -1608,6 +1733,8 @@ export class AssetService {
}
}
const proxyType = (media?.proxyType || null) as 'image' | 'video' | 'audio' | 'pdf' | null
result.push({
id: a.id,
name:
@@ -1619,7 +1746,7 @@ export class AssetService {
type: a.type,
targetType: a.type === AssetType.symlink ? a.target?.type : null,
status: latestVersion.status,
mediaType: latestVersion.mediaType,
proxyType,
latestChildren,
preview,
createdAt: a.createdAt.toISOString(),
@@ -1883,14 +2010,16 @@ export class AssetService {
private async toPreviewInfo(asset: Asset | AssetWithIncludes): Promise<PreviewInfo | null> {
if (!asset.media) return null
const proxyType = (asset.media?.proxyType || null) as 'image' | 'video' | 'audio' | 'pdf' | null
let thumbnailUrl = undefined
if (asset.mediaType?.startsWith('image/') && asset.media.thumbnail?.key) {
if (proxyType === 'image' && asset.media.thumbnail?.key) {
thumbnailUrl = await s3Service.presign(
process.env.S3_BUCKET || 'shumai',
asset.media.thumbnail.key,
'GET',
)
} else if (asset.mediaType?.startsWith('video/') && asset.media.poster?.key) {
} else if (asset.media.poster?.key) {
thumbnailUrl = await s3Service.presign(
process.env.S3_BUCKET || 'shumai',
asset.media.poster.key,
@@ -1908,12 +2037,13 @@ export class AssetService {
}
return {
mediaType: asset.mediaType,
proxyType,
thumbnailUrl,
originalHeight: asset.media.metadata?.originalHeight,
originalWidth: asset.media.metadata?.originalWidth,
spriteUrl,
duration: asset.media.metadata?.duration,
pageCount: asset.media.metadata?.totalFrames,
}
}
@@ -1936,11 +2066,17 @@ export class AssetService {
a.asset.storageKey.key,
'GET',
)
const attachmentProxyType = (a.asset.media?.proxyType || null) as
| 'image'
| 'video'
| 'audio'
| 'pdf'
| null
attachments.push({
id: a.id,
assetId: a.asset.id,
url,
mediaType: a.asset.mediaType,
proxyType: attachmentProxyType,
})
}
}
@@ -1973,6 +2109,14 @@ export class AssetService {
creator.image = await getAvatarUrl(c.creator.image)
}
const completionLastChangedBy: AssetUserInfo | null = c.completionLastChangedBy
? {
id: c.completionLastChangedBy.id,
name: c.completionLastChangedBy.name,
image: await getAvatarUrl(c.completionLastChangedBy.image),
}
: null
return {
id: c.id,
assetId: c.assetId,
@@ -1986,6 +2130,8 @@ export class AssetService {
createdAt: c.createdAt.toISOString(),
updatedAt: c.updatedAt.toISOString(),
sessionId: c.sessionId,
isCompleted: c.isCompleted,
completionLastChangedBy,
}
}
}
+692
View File
@@ -0,0 +1,692 @@
import { prisma } from '@shumai/db'
import { setupTestDbHooks } from '@shumai/db/test'
import { chatService, buildSessionMessages, mapEntryToMessage, type PathEntry } from './chat'
import { describe, expect, it, vi } from 'vitest'
import { workflowService } from '@shumai/workflow-core'
describe('ChatService', () => {
setupTestDbHooks()
async function setupBasicData() {
const user = await prisma.user.create({
data: { name: 'Test User', email: `test-${Date.now()}@example.com`, password: 'pw' },
})
const team = await prisma.team.create({
data: {
name: 'Test Team',
settings: { transcode: { videoStrategy: 'best_match' }, chatbotAgentId: 'test-agent-id' },
sandbox: { create: {} },
},
})
const teamMember = await prisma.teamMember.create({
data: {
teamId: team.id,
userId: user.id,
role: 'owner',
},
})
const project = await prisma.project.create({
data: {
name: 'Test Project',
teamId: team.id,
},
})
const rootFolder = await prisma.asset.create({
data: {
name: 'root',
type: 'root',
status: 'processed',
projectId: project.id,
},
})
await prisma.project.update({
where: { id: project.id },
data: { rootFolderId: rootFolder.id },
})
const agentUser = await prisma.user.create({
data: {
id: 'test-agent-id',
name: 'Test Agent',
email: 'test-agent@shumai.ai',
type: 'agent',
},
})
const agent = await prisma.agent.create({
data: {
id: 'test-agent-id',
teamId: team.id,
type: 'chat',
config: {
provider: 'openai',
model: 'gpt-4',
},
},
})
return { user, team, teamMember, project, rootFolder, agent, agentUser }
}
it('should start a new chat session and trigger workflow', async () => {
const { user, team, project, rootFolder } = await setupBasicData()
const { sessionId, taskId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'hello world',
projectId: project.id,
})
expect(sessionId).toBeDefined()
expect(taskId).toBeDefined()
// Verify AgentSession is created
const session = await prisma.agentSession.findUnique({
where: { id: sessionId },
})
expect(session).toBeDefined()
expect(session?.userId).toBe(user.id)
expect(session?.assetId).toBe(rootFolder.id)
// Verify no comment is created
const comment = await prisma.assetComment.findFirst({
where: { sessionId },
})
expect(comment).toBeNull()
// Verify no session entry is created yet (it is created automatically when the workflow runs)
const entry = await prisma.agentSessionEntry.findFirst({
where: { sessionId },
})
expect(entry).toBeNull()
// Verify WorkflowTask is created with correct payload
const task = await prisma.workflowTask.findUnique({
where: { id: taskId },
})
expect(task).toBeDefined()
expect(task?.type).toBe('chat')
expect(['pending', 'processing']).toContain(task?.status)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const taskPayload = task?.payload as any
expect(taskPayload?.agent?.prompt).toBe('hello world')
})
it('should continue an existing chat session', async () => {
const { user, team, project } = await setupBasicData()
// Start
const { sessionId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'first message',
projectId: project.id,
})
// Continue
const { sessionId: nextSessionId, taskId: nextTaskId } = await chatService.startOrContinueChat(
user,
team.id,
{
agentId: 'test-agent-id',
textPrompt: 'second message',
sessionId,
},
)
expect(nextSessionId).toBe(sessionId)
expect(nextTaskId).toBeDefined()
// Verify no entries are created yet (they are created automatically when the workflow runs)
const entries = await prisma.agentSessionEntry.findMany({
where: { sessionId },
})
expect(entries).toHaveLength(0)
// Verify WorkflowTask is created with correct payload
const task = await prisma.workflowTask.findUnique({
where: { id: nextTaskId },
})
expect(task).toBeDefined()
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const taskPayload = task?.payload as any
expect(taskPayload?.agent?.prompt).toBe('second message')
expect(taskPayload?.agent?.hasAssetChanged).toBe(false)
})
it('should update session assetId and set hasAssetChanged when continuing chat with a new contextAssetId', async () => {
const { user, team, project, rootFolder } = await setupBasicData()
const newFolder = await prisma.asset.create({
data: {
name: 'new_folder',
type: 'folder',
status: 'processed',
projectId: project.id,
parentId: rootFolder.id,
},
})
// Start session on rootFolder
const { sessionId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'message in root folder',
projectId: project.id,
contextAssetId: rootFolder.id,
})
const sessionBefore = await prisma.agentSession.findUnique({ where: { id: sessionId } })
expect(sessionBefore?.assetId).toBe(rootFolder.id)
// Continue session on newFolder
const { taskId: nextTaskId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'message after switching to new folder',
sessionId,
contextAssetId: newFolder.id,
})
// Verify session assetId updated
const sessionAfter = await prisma.agentSession.findUnique({ where: { id: sessionId } })
expect(sessionAfter?.assetId).toBe(newFolder.id)
// Verify task payload has assetId set to newFolder.id and hasAssetChanged is true
const task = await prisma.workflowTask.findUnique({ where: { id: nextTaskId } })
expect(task?.assetId).toBe(newFolder.id)
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const taskPayload = task?.payload as any
expect(taskPayload?.agent?.hasAssetChanged).toBe(true)
})
it('should inject context of referenced assets and attached files', async () => {
const { user, team, project, rootFolder } = await setupBasicData()
const childFolder = await prisma.asset.create({
data: {
name: 'child_folder',
type: 'folder',
status: 'processed',
projectId: project.id,
parentId: rootFolder.id,
},
})
const storageKey = await prisma.storageKey.create({
data: {
key: 'doc_s3_key',
},
})
const textFile = await prisma.asset.create({
data: {
name: 'doc.jpg',
type: 'file',
mediaType: 'image/jpeg',
// eslint-disable-next-line @typescript-eslint/no-explicit-any -- partial MediaInfo mock for testing
media: { proxyType: 'image' } as any,
status: 'processed',
projectId: project.id,
parentId: childFolder.id,
storageKeyId: storageKey.id,
},
})
const { taskId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'process file',
projectId: project.id,
assetIds: [childFolder.id],
attachedFiles: [textFile.id],
})
const task = await prisma.workflowTask.findUnique({
where: { id: taskId },
})
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const taskPayload = task?.payload as any
expect(taskPayload?.agent?.prompt).toBe('process file')
expect(taskPayload?.agent?.attachedFiles).toContain(textFile.id)
expect(taskPayload?.agent?.assetIds).toContain(childFolder.id)
expect(taskPayload?.agent?.imageUrls).toContain('doc_s3_key')
})
it('should list sessions of a user and only return chat type sessions', async () => {
const { user, team, project } = await setupBasicData()
await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'chat 1',
projectId: project.id,
})
await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'chat 2',
projectId: project.id,
})
// Create a comment type session (which should be ignored by listSessions)
const defaultAgent = await prisma.agent.findFirst()
await prisma.agentSession.create({
data: {
agentId: defaultAgent ? defaultAgent.id : 'default',
userId: user.id,
cwd: process.cwd(),
type: 'comment',
},
})
const list = await chatService.listSessions(user.id, team.id, { first: 10 })
expect(list.data).toHaveLength(2)
})
it('should list messages mapped correctly', async () => {
const { user, team, project } = await setupBasicData()
const { sessionId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'hello',
projectId: project.id,
})
// Manually insert a test entry representing user message
await prisma.agentSessionEntry.create({
data: {
id: 'test-entry-id',
sessionId,
entry: {
type: 'message',
id: 'test-entry-id',
parentId: null,
timestamp: new Date().toISOString(),
message: {
role: 'user',
content: [{ type: 'text', text: 'hello' }],
timestamp: Date.now(),
},
},
},
})
const messages = await chatService.listMessages(user.id, team.id, sessionId)
expect(messages).toHaveLength(1)
// Casting to any to access properties of union in test assertions
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const msg = messages[0] as any
expect(msg.role).toBe('user')
expect(msg.content).toEqual([{ type: 'text', text: 'hello' }])
})
it('should hide customType context messages', async () => {
const { user, team, project } = await setupBasicData()
const { sessionId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'hello',
projectId: project.id,
})
// 1. Manually insert a test entry representing user message
await prisma.agentSessionEntry.create({
data: {
id: 'test-entry-1-user',
sessionId,
entry: {
type: 'message',
id: 'test-entry-1-user',
parentId: null,
timestamp: new Date().toISOString(),
message: {
role: 'user',
content: [{ type: 'text', text: 'hello' }],
timestamp: Date.now(),
},
} as unknown as PrismaJson.PiSessionEntry,
},
})
// 2. Manually insert a test entry representing a custom context message
await prisma.agentSessionEntry.create({
data: {
id: 'test-entry-2-context',
sessionId,
entry: {
type: 'custom_message',
id: 'test-entry-2-context',
parentId: 'test-entry-1-user',
timestamp: new Date().toISOString(),
customType: 'context',
content: 'some context',
display: 'chat',
} as unknown as PrismaJson.PiSessionEntry,
},
})
// 3. Manually insert a test entry representing a custom thinking level change
await prisma.agentSessionEntry.create({
data: {
id: 'test-entry-3-thinking',
sessionId,
entry: {
type: 'thinking_level_change',
id: 'test-entry-3-thinking',
parentId: 'test-entry-2-context',
timestamp: new Date().toISOString(),
thinkingLevel: 'deep',
} as unknown as PrismaJson.PiSessionEntry,
},
})
// 4. Manually insert a test entry representing custom context_display_info entry
await prisma.agentSessionEntry.create({
data: {
id: 'test-entry-4-display',
sessionId,
entry: {
type: 'custom',
id: 'test-entry-4-display',
parentId: 'test-entry-3-thinking',
timestamp: new Date().toISOString(),
customType: 'context_display_info',
data: {
assets: [{ id: 'a1', name: 'File A', type: 'file' }],
},
} as unknown as PrismaJson.PiSessionEntry,
},
})
// Verify listMessages returns only the user, thinking level change and custom display entry message
const messages = await chatService.listMessages(user.id, team.id, sessionId)
expect(messages).toHaveLength(3)
expect(messages[0].id).toBe('test-entry-1-user')
expect(messages[1].id).toBe('test-entry-3-thinking')
expect(messages[2].id).toBe('test-entry-4-display')
expect(messages[2].role).toBe('custom')
expect((messages[2] as { customType?: string }).customType).toBe('context_display_info')
// eslint-disable-next-line @typescript-eslint/no-explicit-any
expect((messages[2] as any).details).toEqual({
assets: [{ id: 'a1', name: 'File A', type: 'file' }],
})
// Verify mapEntryToMessage returns null for context message
const contextRecord = await prisma.agentSessionEntry.findUnique({
where: { id: 'test-entry-2-context' },
})
expect(contextRecord).not.toBeNull()
const mapped = mapEntryToMessage(contextRecord!)
expect(mapped).toBeNull()
})
it('should delete session and cascade', async () => {
const { user, team, project } = await setupBasicData()
const { sessionId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'to delete',
projectId: project.id,
})
await chatService.deleteSession(user.id, team.id, sessionId)
const session = await prisma.agentSession.findUnique({
where: { id: sessionId },
})
expect(session).toBeNull()
const entries = await prisma.agentSessionEntry.findMany({
where: { sessionId },
})
expect(entries).toHaveLength(0)
})
it('should build session messages correctly handling compaction and custom messages', () => {
const pathEntries = [
{
id: 'entry-1',
type: 'message',
timestamp: '2026-07-07T00:00:00.000Z',
message: {
role: 'user',
content: 'first message',
},
},
{
id: 'entry-compaction',
type: 'compaction',
timestamp: '2026-07-07T00:01:00.000Z',
summary: 'Compacted conversation history',
tokensBefore: 200,
firstKeptEntryId: 'entry-2',
},
{
id: 'entry-2',
type: 'message',
timestamp: '2026-07-07T00:02:00.000Z',
message: {
role: 'assistant',
content: 'kept message',
},
},
{
id: 'entry-3',
type: 'thinking_level_change',
timestamp: '2026-07-07T00:03:00.000Z',
thinkingLevel: 'high',
},
{
id: 'entry-4',
type: 'branch_summary',
timestamp: '2026-07-07T00:04:00.000Z',
summary: 'Branch summary details',
fromId: 'entry-old',
},
]
const messages = buildSessionMessages(pathEntries as unknown as PathEntry[])
expect(messages).toHaveLength(4)
// 1. Compaction summary message
// Casting to any to access properties of union in test assertions
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const msg0 = messages[0] as any
expect(msg0.role).toBe('custom')
expect(msg0.customType).toBe('compaction-summary')
expect(msg0.content).toBe('Compacted conversation history')
expect((msg0.details as Record<string, unknown>).tokensBefore).toBe(200)
// 2. Kept entry (entry-2)
// Casting to any to access properties of union in test assertions
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const msg1 = messages[1] as any
expect(msg1.id).toBe('entry-2')
expect(msg1.role).toBe('assistant')
expect(msg1.content).toBe('kept message')
// 3. Thinking level change (entry-3)
// Casting to any to access properties of union in test assertions
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const msg2 = messages[2] as any
expect(msg2.role).toBe('thinking_level_change')
expect(msg2.content).toContain('Thinking level changed to high')
// 4. Branch summary (entry-4)
// Casting to any to access properties of union in test assertions
// eslint-disable-next-line @typescript-eslint/no-explicit-any
const msg3 = messages[3] as any
expect(msg3.role).toBe('custom')
expect(msg3.customType).toBe('branch-summary')
expect(msg3.content).toBe('Branch summary details')
expect((msg3.details as Record<string, unknown>).fromId).toBe('entry-old')
})
describe('getNewSessionMessages', () => {
it('returns new messages and lastEntryId', async () => {
const { user, team, project } = await setupBasicData()
const { sessionId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'hello',
projectId: project.id,
})
// Insert two entries
const entry1 = await prisma.agentSessionEntry.create({
data: {
id: 'entry-id-1',
sessionId,
entry: {
type: 'message',
id: 'entry-id-1',
parentId: null,
timestamp: new Date().toISOString(),
message: {
role: 'user',
content: [{ type: 'text', text: 'msg 1' }],
},
} as unknown as PrismaJson.PiSessionEntry,
},
})
const entry2 = await prisma.agentSessionEntry.create({
data: {
id: 'entry-id-2',
sessionId,
entry: {
type: 'message',
id: 'entry-id-2',
parentId: 'entry-id-1',
timestamp: new Date().toISOString(),
message: {
role: 'assistant',
content: [{ type: 'text', text: 'msg 2' }],
},
} as unknown as PrismaJson.PiSessionEntry,
},
})
// 1. Get all messages
const res1 = await chatService.getNewSessionMessages(sessionId)
expect(res1.messages).toHaveLength(2)
expect(res1.lastEntryId).toBe(entry2.id)
expect(res1.messages[0].id).toBe(entry1.id)
// 2. Get with lastEntryId
const res2 = await chatService.getNewSessionMessages(sessionId, entry1.id)
expect(res2.messages).toHaveLength(1)
expect(res2.lastEntryId).toBe(entry2.id)
expect(res2.messages[0].id).toBe(entry2.id)
// 3. Get with latest lastEntryId
const res3 = await chatService.getNewSessionMessages(sessionId, entry2.id)
expect(res3.messages).toHaveLength(0)
expect(res3.lastEntryId).toBe(entry2.id)
})
})
describe('getChatWorkflowStatus', () => {
it('returns task status and output', async () => {
const { team, project } = await setupBasicData()
const rootFolder = await prisma.asset.findFirst({
where: { projectId: project.id, type: 'root' },
})
const task = await prisma.workflowTask.create({
data: {
assetId: rootFolder!.id,
type: 'chat',
status: 'completed',
teamId: team.id,
projectId: project.id,
payload: { projectId: project.id },
output: { success: true },
},
})
const statusRes = await chatService.getChatWorkflowStatus(task.id)
expect(statusRes).not.toBeNull()
expect(statusRes?.status).toBe('completed')
expect(statusRes?.output).toEqual({ success: true })
})
it('returns null if task not found', async () => {
const statusRes = await chatService.getChatWorkflowStatus('non-existent-task')
expect(statusRes).toBeNull()
})
})
describe('abortSession', () => {
it('cancels active task and marks failed/aborted', async () => {
const { user, team, project } = await setupBasicData()
const { sessionId, taskId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'long execution',
projectId: project.id,
})
// Mark the task as processing
await prisma.workflowTask.update({
where: { id: taskId },
data: { status: 'processing' },
})
const cancelSpy = vi.spyOn(workflowService, 'cancel').mockResolvedValue(undefined)
await chatService.abortSession(user.id, sessionId)
expect(cancelSpy).toHaveBeenCalledWith(taskId)
const updatedTask = await prisma.workflowTask.findUnique({
where: { id: taskId },
})
expect(updatedTask?.status).toBe('failed')
expect(updatedTask?.output).toEqual({ error: 'aborted' })
cancelSpy.mockRestore()
})
it('throws error if session not found', async () => {
await expect(chatService.abortSession('user-1', 'non-existent-session')).rejects.toThrow(
'Session not found',
)
})
it('throws error if unauthorized', async () => {
const { team, project } = await setupBasicData()
const otherUser = await prisma.user.create({
data: { name: 'Other User', email: `other-${Date.now()}@example.com`, password: 'pw' },
})
const { sessionId } = await chatService.startOrContinueChat(otherUser, team.id, {
agentId: 'test-agent-id',
textPrompt: 'hello',
projectId: project.id,
})
await expect(chatService.abortSession('different-user-id', sessionId)).rejects.toThrow(
'Unauthorized session access',
)
})
it('throws error if no active task', async () => {
const { user, team, project } = await setupBasicData()
const { sessionId, taskId } = await chatService.startOrContinueChat(user, team.id, {
agentId: 'test-agent-id',
textPrompt: 'hello',
projectId: project.id,
})
// Complete the task
await prisma.workflowTask.update({
where: { id: taskId },
data: { status: 'completed' },
})
await expect(chatService.abortSession(user.id, sessionId)).rejects.toThrow(
'No active execution found for this session',
)
})
})
})
+496
View File
@@ -0,0 +1,496 @@
import { prisma } from '@shumai/db'
import type { Prisma } from '@shumai/db'
import { authzService, Permission, ResourceType } from '@shumai/core/src/authz/authz'
import { paginateQuery, type PaginatedData } from '@shumai/core/src/pagination'
import type { ChatRequest, ChatSessionInfo, ChatMessage } from '@shumai/dtos'
import type { SessionTreeEntry } from '@earendil-works/pi-agent-core'
import { workflowService } from '@shumai/workflow-core'
type User = Prisma.UserGetPayload<Record<string, never>>
export type PathEntry = SessionTreeEntry & { id: string }
export function buildSessionMessages(pathEntries: PathEntry[]): ChatMessage[] {
let compaction: (SessionTreeEntry & { type: 'compaction' } & { id: string }) | null = null
for (const entry of pathEntries) {
if (entry.type === 'compaction') {
compaction = entry as SessionTreeEntry & { type: 'compaction' } & { id: string }
}
}
const messages: ChatMessage[] = []
const appendMessage = (entry: PathEntry) => {
const timestampMs = new Date(entry.timestamp).getTime()
if (entry.type === 'message') {
messages.push({
...entry.message,
id: entry.id,
timestamp: entry.message.timestamp || timestampMs,
} as unknown as ChatMessage)
} else if (entry.type === 'custom_message') {
if (entry.customType === 'context') return
messages.push({
id: entry.id,
role: 'custom',
customType: entry.customType,
content: entry.content,
display: entry.display,
details: entry.details as Record<string, unknown> | undefined,
timestamp: timestampMs,
} as unknown as ChatMessage)
} else if (entry.type === 'custom') {
messages.push({
id: entry.id,
role: 'custom',
customType: entry.customType,
details: (entry as { data?: unknown }).data as Record<string, unknown> | undefined,
timestamp: timestampMs,
} as unknown as ChatMessage)
} else if (entry.type === 'branch_summary') {
messages.push({
id: entry.id,
role: 'custom',
customType: 'branch-summary',
content: entry.summary,
details: { fromId: entry.fromId },
timestamp: timestampMs,
} as unknown as ChatMessage)
} else if (entry.type === 'thinking_level_change') {
messages.push({
id: entry.id,
role: 'thinking_level_change',
content: `Thinking level changed to ${entry.thinkingLevel}`,
timestamp: timestampMs,
} as unknown as ChatMessage)
}
}
if (compaction) {
const compactionTimestamp = new Date(compaction.timestamp).getTime()
messages.push({
id: compaction.id,
role: 'custom',
customType: 'compaction-summary',
content: compaction.summary,
details: { tokensBefore: compaction.tokensBefore },
timestamp: compactionTimestamp,
} as unknown as ChatMessage)
const compactionIdx = pathEntries.findIndex(
(e) => e.type === 'compaction' && e.id === compaction?.id,
)
let foundFirstKept = false
for (let i = 0; i < compactionIdx; i++) {
const entry = pathEntries[i]
if (entry.id === compaction.firstKeptEntryId) foundFirstKept = true
if (foundFirstKept) appendMessage(entry)
}
for (let i = compactionIdx + 1; i < pathEntries.length; i++) {
appendMessage(pathEntries[i])
}
} else {
for (const entry of pathEntries) {
appendMessage(entry)
}
}
return messages
}
export function mapEntryToMessage(
entryRecord: Prisma.AgentSessionEntryGetPayload<Record<string, never>>,
): ChatMessage | null {
const entryObj = entryRecord.entry as unknown as SessionTreeEntry
if (
entryObj.type === 'custom_message' &&
(entryObj as { customType?: string }).customType === 'context'
) {
return null
}
const pathEntries = [{ ...entryObj, id: entryRecord.id }]
const messages = buildSessionMessages(pathEntries)
if (messages.length > 0) {
return messages[0]
}
return {
id: entryRecord.id,
role: 'user',
content: '',
} as unknown as ChatMessage
}
export class ChatService {
constructor(private readonly prismaClient: typeof prisma = prisma) {}
private async getAssetPath(assetId: string): Promise<string> {
const parts: string[] = []
let currentId: string | null = assetId
while (currentId) {
const dbAssetNode = (await prisma.asset.findUnique({
where: { id: currentId },
select: { id: true, name: true, parentId: true, type: true },
})) as { id: string; name: string; parentId: string | null; type: unknown } | null
if (!dbAssetNode) break
if (dbAssetNode.type === 'root' || dbAssetNode.type === 'share_root') {
break
}
parts.unshift(dbAssetNode.name)
currentId = dbAssetNode.parentId
}
return parts.join('/')
}
async startOrContinueChat(
user: User,
teamId: string,
req: ChatRequest,
): Promise<{ sessionId: string; taskId: string }> {
const {
agentId,
textPrompt,
attachedFiles = [],
assetIds = [],
sessionId: passedSessionId,
contextAssetId,
projectId: passedProjectId,
} = req
// 1. Permissions verification
if (contextAssetId) {
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Asset,
id: contextAssetId,
})
}
for (const fileId of attachedFiles) {
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Asset,
id: fileId,
})
}
for (const assetId of assetIds) {
await authzService.hasPermission({
user,
permission: Permission.Read,
type: ResourceType.Asset,
id: assetId,
})
}
// 2. Project / Asset scoping resolution
let targetAssetId = ''
let hasAssetChanged = false
if (passedSessionId) {
const session = await this.prismaClient.agentSession.findUnique({
where: { id: passedSessionId },
include: { asset: { include: { project: true } } },
})
if (!session) throw new Error('Session not found')
if (session.userId !== user.id) throw new Error('Unauthorized session access')
if (!session.assetId) throw new Error('Session has no associated asset')
if (contextAssetId && contextAssetId !== session.assetId) {
const contextAsset = await this.prismaClient.asset.findUnique({
where: { id: contextAssetId },
include: { project: true },
})
if (!contextAsset) throw new Error('Context asset not found')
targetAssetId = contextAssetId
hasAssetChanged = true
} else {
targetAssetId = session.assetId
}
} else {
// Start a new session
if (contextAssetId) {
const asset = await this.prismaClient.asset.findUnique({
where: { id: contextAssetId },
include: { project: true },
})
if (!asset) throw new Error('Context asset not found')
targetAssetId = contextAssetId
} else if (passedProjectId) {
const project = await this.prismaClient.project.findUnique({
where: { id: passedProjectId },
})
if (!project || !project.rootFolderId) throw new Error('Project or root folder not found')
targetAssetId = project.rootFolderId
} else {
// Find default project from user's first team
const teamMember = await this.prismaClient.teamMember.findFirst({
where: { userId: user.id },
include: { team: { include: { projects: { include: { rootFolder: true } } } } },
})
const project = teamMember?.team?.projects?.[0]
if (!project || !project.rootFolderId) {
throw new Error('No active project found for the user. Please specify a projectId.')
}
targetAssetId = project.rootFolderId
}
}
const finalPrompt = textPrompt || ''
const resolvedImageUrls: string[] = []
for (const fileId of attachedFiles) {
const file = await this.prismaClient.asset.findUnique({
where: { id: fileId },
include: { storageKey: true },
})
const isImage = (file?.media as PrismaJson.MediaInfo | null)?.proxyType === 'image'
if (file && isImage && file.storageKey?.key) {
resolvedImageUrls.push(file.storageKey.key)
}
}
// 4. Create database records in a transaction
return await this.prismaClient.$transaction(async (tx) => {
const asset = await tx.asset.findUnique({
where: { id: targetAssetId },
include: { project: { include: { team: true } } },
})
if (!asset || !asset.project) {
throw new Error('Resolved asset or project not found')
}
const teamId = asset.project.teamId
const projectId = asset.project.id
let activeSessionId = passedSessionId
// Ensure AI agent configuration exists
const agent = await tx.agent.findUnique({ where: { id: agentId } })
if (!agent) {
throw new Error(`Agent with ID "${agentId}" not found`)
}
if (agent.teamId !== teamId) {
throw new Error(`Agent does not belong to the specified team`)
}
if (activeSessionId) {
const sessionExists = await tx.agentSession.findUnique({
where: { id: activeSessionId },
select: { id: true, userId: true },
})
if (!sessionExists) throw new Error('Session not found')
if (sessionExists.userId !== user.id) throw new Error('Unauthorized session access')
// Update the session's agentId and assetId (if location changed)
await tx.agentSession.update({
where: { id: activeSessionId },
data: {
agentId,
...(hasAssetChanged ? { assetId: targetAssetId } : {}),
},
})
} else {
const newSession = await tx.agentSession.create({
data: {
agentId,
userId: user.id,
cwd: process.cwd(),
assetId: asset.id,
type: 'chat',
},
})
activeSessionId = newSession.id
}
// Trigger chat workflow
const task = await tx.workflowTask.create({
data: {
assetId: asset.id,
type: 'chat',
status: 'pending',
teamId,
projectId,
sessionId: activeSessionId,
payload: {
projectId,
agent: {
prompt: finalPrompt,
imageUrls: resolvedImageUrls,
attachedFiles,
assetIds,
explicitMention: true,
agentId,
sessionId: activeSessionId,
userId: user.id,
isNewChat: !passedSessionId,
hasAssetChanged,
},
},
},
})
return {
sessionId: activeSessionId!,
taskId: task.id,
}
})
}
async listSessions(
userId: string,
teamId: string,
params: { first?: number; after?: string },
): Promise<PaginatedData<ChatSessionInfo[]>> {
const where: Prisma.AgentSessionWhereInput = {
userId,
type: 'chat',
agent: {
teamId,
},
}
return await paginateQuery(
async (skip, take) => {
const sessions = await this.prismaClient.agentSession.findMany({
where,
skip,
take,
orderBy: { id: 'desc' },
})
return sessions.map((s) => ({
id: s.id,
agentId: s.agentId,
userId: s.userId,
assetId: s.assetId,
userCommentId: s.userCommentId,
name: s.name,
createdAt: s.createdAt.toISOString(),
updatedAt: s.updatedAt.toISOString(),
}))
},
async () => this.prismaClient.agentSession.count({ where }),
params,
)
}
async listMessages(userId: string, teamId: string, sessionId: string): Promise<ChatMessage[]> {
const session = await this.prismaClient.agentSession.findUnique({
where: { id: sessionId },
})
if (!session) {
throw new Error('Session not found')
}
if (session.userId !== userId) {
throw new Error('Unauthorized session access')
}
const entries = await this.prismaClient.agentSessionEntry.findMany({
where: { sessionId },
orderBy: { id: 'asc' },
})
const pathEntries = entries.map((e) => {
const entryObj = e.entry as unknown as SessionTreeEntry
return {
...entryObj,
id: e.id,
}
})
return buildSessionMessages(pathEntries)
}
async deleteSession(userId: string, teamId: string, sessionId: string): Promise<void> {
const session = await this.prismaClient.agentSession.findUnique({
where: { id: sessionId },
})
if (!session) {
throw new Error('Session not found')
}
if (session.userId !== userId) {
throw new Error('Unauthorized session access')
}
await this.prismaClient.agentSession.delete({
where: { id: sessionId },
})
}
async getNewSessionMessages(
sessionId: string,
lastEntryId?: string,
): Promise<{ messages: ChatMessage[]; lastEntryId: string | null }> {
const entries = await this.prismaClient.agentSessionEntry.findMany({
where: {
sessionId,
...(lastEntryId ? { id: { gt: lastEntryId } } : {}),
},
orderBy: { id: 'asc' },
})
if (entries.length === 0) {
return { messages: [], lastEntryId: lastEntryId ?? null }
}
const messages: ChatMessage[] = []
for (const entry of entries) {
const message = mapEntryToMessage(entry)
if (message) {
messages.push(message)
}
}
return {
messages,
lastEntryId: entries[entries.length - 1].id,
}
}
async getChatWorkflowStatus(
taskId: string,
): Promise<{ status: string | null; output: Record<string, unknown> | null } | null> {
const task = await this.prismaClient.workflowTask.findUnique({
where: { id: taskId },
select: { status: true, output: true },
})
if (!task) return null
return {
status: task.status,
output: task.output as Record<string, unknown> | null,
}
}
async abortSession(userId: string, sessionId: string): Promise<void> {
const session = await this.prismaClient.agentSession.findUnique({
where: { id: sessionId },
})
if (!session) {
throw new Error('Session not found')
}
if (session.userId !== userId) {
throw new Error('Unauthorized session access')
}
const taskToAbort = await this.prismaClient.workflowTask.findFirst({
where: {
status: { in: ['pending', 'processing'] },
type: 'chat',
sessionId,
},
})
if (!taskToAbort) {
throw new Error('No active execution found for this session')
}
await workflowService.cancel(taskToAbort.id)
await this.prismaClient.workflowTask.update({
where: { id: taskToAbort.id },
data: {
status: 'failed',
output: { error: 'aborted' },
},
})
}
}
export const chatService = new ChatService()
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,65 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'ant-ling'
export const providerConfig = {
name: 'ant-ling',
config: {
api: 'openai-completions',
baseUrl: 'https://api.ant-ling.com/v1',
apiKey: '',
},
models: [
{
modelId: 'Ling-2.6-1T',
name: 'Ling 2.6 1T',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.06,
output: 0.25,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Ling-2.6-flash',
name: 'Ling 2.6 Flash',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.01,
output: 0.02,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Ring-2.6-1T',
name: 'Ring 2.6 1T',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.06,
output: 0.25,
cacheRead: 0,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,252 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'anthropic'
export const providerConfig = {
name: 'anthropic',
config: {
api: 'anthropic-messages',
baseUrl: 'https://api.anthropic.com',
apiKey: 'ANTHROPIC_API_KEY',
},
models: [
{
modelId: 'claude-fable-5',
name: 'Claude Fable 5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 10,
output: 50,
cacheRead: 1,
cacheWrite: 12.5,
},
},
},
{
modelId: 'claude-haiku-4-5',
name: 'Claude Haiku 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 1,
output: 5,
cacheRead: 0.1,
cacheWrite: 1.25,
},
},
},
{
modelId: 'claude-haiku-4-5-20251001',
name: 'Claude Haiku 4.5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 1,
output: 5,
cacheRead: 0.1,
cacheWrite: 1.25,
},
},
},
{
modelId: 'claude-opus-4-1',
name: 'Claude Opus 4.1 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 15,
output: 75,
cacheRead: 1.5,
cacheWrite: 18.75,
},
},
},
{
modelId: 'claude-opus-4-1-20250805',
name: 'Claude Opus 4.1',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 15,
output: 75,
cacheRead: 1.5,
cacheWrite: 18.75,
},
},
},
{
modelId: 'claude-opus-4-5',
name: 'Claude Opus 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-5-20251101',
name: 'Claude Opus 4.5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-6',
name: 'Claude Opus 4.6',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-7',
name: 'Claude Opus 4.7',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-8',
name: 'Claude Opus 4.8',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-sonnet-4-5',
name: 'Claude Sonnet 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 64000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-4-5-20250929',
name: 'Claude Sonnet 4.5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 64000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-4-6',
name: 'Claude Sonnet 4.6',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-5',
name: 'Claude Sonnet 5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 2,
output: 10,
cacheRead: 0.2,
cacheWrite: 2.5,
},
},
},
],
} as const
@@ -0,0 +1,796 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'azure-openai-responses'
export const providerConfig = {
name: 'azure-openai-responses',
config: {
api: 'azure-openai-responses',
baseUrl: '',
apiKey: 'AZURE_OPENAI_API_KEY',
},
models: [
{
modelId: 'gpt-4',
name: 'GPT-4',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text'],
contextWindow: 8192,
maxTokens: 8192,
cost: {
input: 30,
output: 60,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4-turbo',
name: 'GPT-4 Turbo',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 4096,
cost: {
input: 10,
output: 30,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4.1',
name: 'GPT-4.1',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 1047576,
maxTokens: 32768,
cost: {
input: 2,
output: 8,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4.1-mini',
name: 'GPT-4.1 mini',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 1047576,
maxTokens: 32768,
cost: {
input: 0.4,
output: 1.6,
cacheRead: 0.1,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4.1-nano',
name: 'GPT-4.1 nano',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 1047576,
maxTokens: 32768,
cost: {
input: 0.1,
output: 0.4,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o',
name: 'GPT-4o',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2.5,
output: 10,
cacheRead: 1.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-2024-05-13',
name: 'GPT-4o (2024-05-13)',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 4096,
cost: {
input: 5,
output: 15,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-2024-08-06',
name: 'GPT-4o (2024-08-06)',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2.5,
output: 10,
cacheRead: 1.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-2024-11-20',
name: 'GPT-4o (2024-11-20)',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2.5,
output: 10,
cacheRead: 1.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-mini',
name: 'GPT-4o mini',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 0.15,
output: 0.6,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5',
name: 'GPT-5',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-chat-latest',
name: 'GPT-5 Chat Latest',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-codex',
name: 'GPT-5-Codex',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-mini',
name: 'GPT-5 Mini',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.25,
output: 2,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-nano',
name: 'GPT-5 Nano',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.05,
output: 0.4,
cacheRead: 0.005,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-pro',
name: 'GPT-5 Pro',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 15,
output: 120,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1',
name: 'GPT-5.1',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-chat-latest',
name: 'GPT-5.1 Chat',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex',
name: 'GPT-5.1 Codex',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex-max',
name: 'GPT-5.1 Codex Max',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex-mini',
name: 'GPT-5.1 Codex mini',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.25,
output: 2,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2',
name: 'GPT-5.2',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-chat-latest',
name: 'GPT-5.2 Chat',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-codex',
name: 'GPT-5.2 Codex',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-pro',
name: 'GPT-5.2 Pro',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 21,
output: 168,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-chat-latest',
name: 'GPT-5.3 Chat (latest)',
config: {
api: 'azure-openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-codex',
name: 'GPT-5.3 Codex',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-codex-spark',
name: 'GPT-5.3 Codex Spark',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 32000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4',
name: 'GPT-5.4',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-mini',
name: 'GPT-5.4 mini',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.75,
output: 4.5,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-nano',
name: 'GPT-5.4 nano',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.2,
output: 1.25,
cacheRead: 0.02,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-pro',
name: 'GPT-5.4 Pro',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 30,
output: 180,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.5',
name: 'GPT-5.5',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.5-pro',
name: 'GPT-5.5 Pro',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 30,
output: 180,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.6-luna',
name: 'GPT-5.6 Luna',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 1,
output: 6,
cacheRead: 0.1,
cacheWrite: 1.25,
},
},
},
{
modelId: 'gpt-5.6-sol',
name: 'GPT-5.6 Sol',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'gpt-5.6-terra',
name: 'GPT-5.6 Terra',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 3.125,
},
},
},
{
modelId: 'gpt-realtime-2.1',
name: 'GPT-Realtime-2.1',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 32000,
cost: {
input: 4,
output: 24,
cacheRead: 0.4,
cacheWrite: 0,
},
},
},
{
modelId: 'o1',
name: 'o1',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 15,
output: 60,
cacheRead: 7.5,
cacheWrite: 0,
},
},
},
{
modelId: 'o1-pro',
name: 'o1-pro',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 150,
output: 600,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'o3',
name: 'o3',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 2,
output: 8,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'o3-deep-research',
name: 'o3-deep-research',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 10,
output: 40,
cacheRead: 2.5,
cacheWrite: 0,
},
},
},
{
modelId: 'o3-mini',
name: 'o3-mini',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 1.1,
output: 4.4,
cacheRead: 0.55,
cacheWrite: 0,
},
},
},
{
modelId: 'o3-pro',
name: 'o3-pro',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 20,
output: 80,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'o4-mini',
name: 'o4-mini',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 1.1,
output: 4.4,
cacheRead: 0.275,
cacheWrite: 0,
},
},
},
{
modelId: 'o4-mini-deep-research',
name: 'o4-mini-deep-research',
config: {
api: 'azure-openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 2,
output: 8,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,65 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'cerebras'
export const providerConfig = {
name: 'cerebras',
config: {
api: 'openai-completions',
baseUrl: 'https://api.cerebras.ai/v1',
apiKey: 'CEREBRAS_API_KEY',
},
models: [
{
modelId: 'gemma-4-31b',
name: 'Gemma 4 31B IT',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 131072,
maxTokens: 40960,
cost: {
input: 0.99,
output: 1.49,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-oss-120b',
name: 'GPT OSS 120B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 40960,
cost: {
input: 0.35,
output: 0.75,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-glm-4.7',
name: 'Z.AI GLM-4.7',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 40960,
cost: {
input: 2.25,
output: 2.75,
cacheRead: 2.25,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,729 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'cloudflare-ai-gateway'
export const providerConfig = {
name: 'cloudflare-ai-gateway',
config: {
api: 'anthropic-messages',
baseUrl:
'https://gateway.ai.cloudflare.com/v1/{CLOUDFLARE_ACCOUNT_ID}/{CLOUDFLARE_GATEWAY_ID}/anthropic',
apiKey: 'CLOUDFLARE_API_KEY',
},
models: [
{
modelId: 'claude-3-5-haiku',
name: 'Claude Haiku 3.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: false,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 8192,
cost: {
input: 0.8,
output: 4,
cacheRead: 0.08,
cacheWrite: 1,
},
},
},
{
modelId: 'claude-3-haiku',
name: 'Claude Haiku 3',
config: {
api: 'anthropic-messages',
reasoning: false,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 4096,
cost: {
input: 0.25,
output: 1.25,
cacheRead: 0.03,
cacheWrite: 0.3,
},
},
},
{
modelId: 'claude-3-opus',
name: 'Claude Opus 3',
config: {
api: 'anthropic-messages',
reasoning: false,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 4096,
cost: {
input: 15,
output: 75,
cacheRead: 1.5,
cacheWrite: 18.75,
},
},
},
{
modelId: 'claude-3-sonnet',
name: 'Claude Sonnet 3',
config: {
api: 'anthropic-messages',
reasoning: false,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 4096,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 0.3,
},
},
},
{
modelId: 'claude-3.5-haiku',
name: 'Claude Haiku 3.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: false,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 8192,
cost: {
input: 0.8,
output: 4,
cacheRead: 0.08,
cacheWrite: 1,
},
},
},
{
modelId: 'claude-3.5-sonnet',
name: 'Claude Sonnet 3.5 v2',
config: {
api: 'anthropic-messages',
reasoning: false,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 8192,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-fable-5',
name: 'Claude Fable 5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 10,
output: 50,
cacheRead: 1,
cacheWrite: 12.5,
},
},
},
{
modelId: 'claude-haiku-4-5',
name: 'Claude Haiku 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 1,
output: 5,
cacheRead: 0.1,
cacheWrite: 1.25,
},
},
},
{
modelId: 'claude-opus-4',
name: 'Claude Opus 4 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 15,
output: 75,
cacheRead: 1.5,
cacheWrite: 18.75,
},
},
},
{
modelId: 'claude-opus-4-1',
name: 'Claude Opus 4.1 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 15,
output: 75,
cacheRead: 1.5,
cacheWrite: 18.75,
},
},
},
{
modelId: 'claude-opus-4-5',
name: 'Claude Opus 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-6',
name: 'Claude Opus 4.6 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-7',
name: 'Claude Opus 4.7',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-8',
name: 'Claude Opus 4.8',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-sonnet-4',
name: 'Claude Sonnet 4 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-4-5',
name: 'Claude Sonnet 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-4-6',
name: 'Claude Sonnet 4.6',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 64000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-5',
name: 'Claude Sonnet 5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 2,
output: 10,
cacheRead: 0.2,
cacheWrite: 2.5,
},
},
},
{
modelId: 'gpt-4',
name: 'GPT-4',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text'],
contextWindow: 8192,
maxTokens: 8192,
cost: {
input: 30,
output: 60,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4-turbo',
name: 'GPT-4 Turbo',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 4096,
cost: {
input: 10,
output: 30,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o',
name: 'GPT-4o',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2.5,
output: 10,
cacheRead: 1.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-mini',
name: 'GPT-4o mini',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 0.15,
output: 0.6,
cacheRead: 0.08,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1',
name: 'GPT-5.1',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.13,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex',
name: 'GPT-5.1 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2',
name: 'GPT-5.2',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-codex',
name: 'GPT-5.2 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-codex',
name: 'GPT-5.3 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4',
name: 'GPT-5.4',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.5',
name: 'GPT-5.5',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.6-luna',
name: 'GPT-5.6 Luna',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 1,
output: 6,
cacheRead: 0.1,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.6-sol',
name: 'GPT-5.6 Sol',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.6-terra',
name: 'GPT-5.6 Terra',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 0,
},
},
},
{
modelId: 'o1',
name: 'o1',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 15,
output: 60,
cacheRead: 7.5,
cacheWrite: 0,
},
},
},
{
modelId: 'o3',
name: 'o3',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 2,
output: 8,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'o3-mini',
name: 'o3-mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 1.1,
output: 4.4,
cacheRead: 0.55,
cacheWrite: 0,
},
},
},
{
modelId: 'o3-pro',
name: 'o3-pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 20,
output: 80,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'o4-mini',
name: 'o4-mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 1.1,
output: 4.4,
cacheRead: 0.28,
cacheWrite: 0,
},
},
},
{
modelId: 'workers-ai/@cf/moonshotai/kimi-k2.5',
name: 'Kimi K2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 256000,
cost: {
input: 0.6,
output: 3,
cacheRead: 0.1,
cacheWrite: 0,
},
},
},
{
modelId: 'workers-ai/@cf/moonshotai/kimi-k2.6',
name: 'Kimi K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 256000,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.16,
cacheWrite: 0,
},
},
},
{
modelId: 'workers-ai/@cf/nvidia/nemotron-3-120b-a12b',
name: 'Nemotron 3 Super 120B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 256000,
maxTokens: 256000,
cost: {
input: 0.5,
output: 1.5,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'workers-ai/@cf/zai-org/glm-4.7-flash',
name: 'GLM-4.7-Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 131072,
cost: {
input: 0.06,
output: 0.4,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'workers-ai/@cf/zai-org/glm-5.2',
name: 'Glm 5.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0.26,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,235 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'cloudflare-workers-ai'
export const providerConfig = {
name: 'cloudflare-workers-ai',
config: {
api: 'openai-completions',
baseUrl: 'https://api.cloudflare.com/client/v4/accounts/{CLOUDFLARE_ACCOUNT_ID}/ai/v1',
apiKey: 'CLOUDFLARE_API_KEY',
},
models: [
{
modelId: '@cf/google/gemma-4-26b-a4b-it',
name: 'Gemma 4 26B A4B IT',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 16384,
cost: {
input: 0.1,
output: 0.3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/ibm-granite/granite-4.0-h-micro',
name: 'Granite 4.0 H Micro',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 131000,
maxTokens: 131000,
cost: {
input: 0.017,
output: 0.112,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/meta/llama-3.3-70b-instruct-fp8-fast',
name: 'Llama 3.3 70B Instruct fp8 Fast',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 24000,
maxTokens: 24000,
cost: {
input: 0.293,
output: 2.253,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/meta/llama-4-scout-17b-16e-instruct',
name: 'Llama 4 Scout 17B 16E Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text', 'image'],
contextWindow: 131000,
maxTokens: 16384,
cost: {
input: 0.27,
output: 0.85,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/mistralai/mistral-small-3.1-24b-instruct',
name: 'Mistral Small 3.1 24B Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.351,
output: 0.555,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/moonshotai/kimi-k2.6',
name: 'Kimi K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 256000,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.16,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/moonshotai/kimi-k2.7-code',
name: 'Kimi K2.7 Code',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.19,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/nvidia/nemotron-3-120b-a12b',
name: 'Nemotron 3 Super 120B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 256000,
maxTokens: 256000,
cost: {
input: 0.5,
output: 1.5,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/openai/gpt-oss-120b',
name: 'GPT OSS 120B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 0.35,
output: 0.75,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/openai/gpt-oss-20b',
name: 'GPT OSS 20B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 0.2,
output: 0.3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/qwen/qwen3-30b-a3b-fp8',
name: 'Qwen3 30B A3b fp8',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 32768,
maxTokens: 32768,
cost: {
input: 0.0509,
output: 0.335,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/zai-org/glm-4.7-flash',
name: 'GLM-4.7-Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 131072,
cost: {
input: 0.0605,
output: 0.4,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: '@cf/zai-org/glm-5.2',
name: 'Glm 5.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0.26,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,48 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'deepseek'
export const providerConfig = {
name: 'deepseek',
config: {
api: 'openai-completions',
baseUrl: 'https://api.deepseek.com',
apiKey: 'DEEPSEEK_API_KEY',
},
models: [
{
modelId: 'deepseek-v4-flash',
name: 'DeepSeek V4 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 0.14,
output: 0.28,
cacheRead: 0.0028,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-v4-pro',
name: 'DeepSeek V4 Pro',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 0.435,
output: 0.87,
cacheRead: 0.003625,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,286 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'fireworks'
export const providerConfig = {
name: 'fireworks',
config: {
api: 'anthropic-messages',
baseUrl: 'https://api.fireworks.ai/inference',
apiKey: 'FIREWORKS_API_KEY',
},
models: [
{
modelId: 'accounts/fireworks/models/deepseek-v4-flash',
name: 'DeepSeek V4 Flash',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 0.14,
output: 0.28,
cacheRead: 0.028,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/deepseek-v4-pro',
name: 'DeepSeek V4 Pro',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 1.74,
output: 3.48,
cacheRead: 0.145,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/glm-5p1',
name: 'GLM 5.1',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 202800,
maxTokens: 131072,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0.26,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/glm-5p2',
name: 'GLM 5.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1048575,
maxTokens: 131072,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0.14,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/gpt-oss-120b',
name: 'GPT OSS 120B',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 32768,
cost: {
input: 0.15,
output: 0.6,
cacheRead: 0.015,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/gpt-oss-20b',
name: 'GPT OSS 20B',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 32768,
cost: {
input: 0.07,
output: 0.3,
cacheRead: 0.035,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/kimi-k2p6',
name: 'Kimi K2.6',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262000,
maxTokens: 262000,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.16,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/kimi-k2p7-code',
name: 'Kimi K2.7 Code',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262000,
maxTokens: 262000,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.19,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/minimax-m2p7',
name: 'MiniMax-M2.7',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 196608,
maxTokens: 196608,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/minimax-m3',
name: 'MiniMax-M3',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 512000,
maxTokens: 512000,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/models/qwen3p7-plus',
name: 'Qwen 3.7 Plus',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.4,
output: 1.6,
cacheRead: 0.08,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/routers/glm-5p1-fast',
name: 'GLM 5.1 Fast',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 202800,
maxTokens: 131072,
cost: {
input: 2.8,
output: 8.8,
cacheRead: 0.52,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/routers/glm-5p2-fast',
name: 'GLM 5.2 Fast',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1048575,
maxTokens: 131072,
cost: {
input: 2.1,
output: 6.6,
cacheRead: 0.21,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/routers/kimi-k2p6-fast',
name: 'Kimi K2.6 Fast',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262000,
maxTokens: 262000,
cost: {
input: 2,
output: 8,
cacheRead: 0.3,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/routers/kimi-k2p6-turbo',
name: 'Kimi K2.6 Turbo',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262000,
maxTokens: 262000,
cost: {
input: 2,
output: 8,
cacheRead: 0.3,
cacheWrite: 0,
},
},
},
{
modelId: 'accounts/fireworks/routers/kimi-k2p7-code-fast',
name: 'Kimi K2.7 Code Fast',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262000,
maxTokens: 262000,
cost: {
input: 1.9,
output: 8,
cacheRead: 0.38,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,544 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'github-copilot'
export const providerConfig = {
name: 'github-copilot',
config: {
api: 'openai-completions',
baseUrl: 'https://api.individual.githubcopilot.com',
apiKey: 'COPILOT_GITHUB_TOKEN',
},
models: [
{
modelId: 'claude-fable-5',
name: 'Claude Fable 5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 10,
output: 50,
cacheRead: 1,
cacheWrite: 12.5,
},
},
},
{
modelId: 'claude-haiku-4.5',
name: 'Claude Haiku 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 1,
output: 5,
cacheRead: 0.1,
cacheWrite: 1.25,
},
},
},
{
modelId: 'claude-opus-4.5',
name: 'Claude Opus 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4.6',
name: 'Claude Opus 4.6',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 32000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4.7',
name: 'Claude Opus 4.7',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 32000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4.8',
name: 'Claude Opus 4.8',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 64000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-sonnet-4',
name: 'Claude Sonnet 4 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 216000,
maxTokens: 16000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-4.5',
name: 'Claude Sonnet 4.5 (latest)',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-4.6',
name: 'Claude Sonnet 4.6',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 32000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-5',
name: 'Claude Sonnet 5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 2,
output: 10,
cacheRead: 0.2,
cacheWrite: 2.5,
},
},
},
{
modelId: 'gemini-2.5-pro',
name: 'Gemini 2.5 Pro',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 64000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3-flash-preview',
name: 'Gemini 3 Flash Preview',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 64000,
cost: {
input: 0.5,
output: 3,
cacheRead: 0.05,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-pro-preview',
name: 'Gemini 3.1 Pro Preview',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 64000,
cost: {
input: 2,
output: 12,
cacheRead: 0.2,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 200000,
input: 4,
output: 18,
cacheRead: 0.4,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gemini-3.5-flash',
name: 'Gemini 3.5 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 1.5,
output: 9,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4.1',
name: 'GPT-4.1',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2,
output: 8,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-mini',
name: 'GPT-5 Mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 264000,
maxTokens: 64000,
cost: {
input: 0.25,
output: 2,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2',
name: 'GPT-5.2',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-codex',
name: 'GPT-5.2 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-codex',
name: 'GPT-5.3 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4',
name: 'GPT-5.4',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 272000,
input: 5,
output: 22.5,
cacheRead: 0.5,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gpt-5.4-mini',
name: 'GPT-5.4 mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.75,
output: 4.5,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-nano',
name: 'GPT-5.4 nano',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.2,
output: 1.25,
cacheRead: 0.02,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.5',
name: 'GPT-5.5',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 272000,
input: 10,
output: 45,
cacheRead: 1,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gpt-5.6-luna',
name: 'GPT-5.6 Luna',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 1,
output: 6,
cacheRead: 0.1,
cacheWrite: 1.25,
tiers: [
{
inputTokensAbove: 200000,
input: 2,
output: 9,
cacheRead: 0.2,
cacheWrite: 2.5,
},
],
},
},
},
{
modelId: 'gpt-5.6-sol',
name: 'GPT-5.6 Sol',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 6.25,
tiers: [
{
inputTokensAbove: 272000,
input: 10,
output: 45,
cacheRead: 1,
cacheWrite: 12.5,
},
],
},
},
},
{
modelId: 'gpt-5.6-terra',
name: 'GPT-5.6 Terra',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 3.125,
tiers: [
{
inputTokensAbove: 272000,
input: 5,
output: 22.5,
cacheRead: 0.5,
cacheWrite: 6.25,
},
],
},
},
},
{
modelId: 'kimi-k2.7-code',
name: 'Kimi K2.7 Code',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 32000,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.19,
cacheWrite: 0,
},
},
},
{
modelId: 'mai-code-1-flash-picker',
name: 'MAI-Code-1-Flash',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text'],
contextWindow: 256000,
maxTokens: 128000,
cost: {
input: 0.75,
output: 4.5,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,184 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'google-vertex'
export const providerConfig = {
name: 'google-vertex',
config: {
api: 'google-vertex',
baseUrl: 'https://{location}-aiplatform.googleapis.com',
apiKey: 'GOOGLE_CLOUD_API_KEY',
},
models: [
{
modelId: 'gemini-2.5-flash',
name: 'Gemini 2.5 Flash',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.3,
output: 2.5,
cacheRead: 0.03,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-2.5-flash-lite',
name: 'Gemini 2.5 Flash-Lite',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.1,
output: 0.4,
cacheRead: 0.01,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-2.5-pro',
name: 'Gemini 2.5 Pro',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3-flash-preview',
name: 'Gemini 3 Flash Preview',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.5,
output: 3,
cacheRead: 0.05,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-flash-lite',
name: 'Gemini 3.1 Flash Lite',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.25,
output: 1.5,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-pro-preview',
name: 'Gemini 3.1 Pro Preview',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 2,
output: 12,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-pro-preview-customtools',
name: 'Gemini 3.1 Pro Preview Custom Tools',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 2,
output: 12,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.5-flash',
name: 'Gemini 3.5 Flash',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.5,
output: 9,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-flash-latest',
name: 'Gemini Flash Latest',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.5,
output: 9,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-flash-lite-latest',
name: 'Gemini Flash-Lite Latest',
config: {
api: 'google-vertex',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.25,
output: 1.5,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,320 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'google'
export const providerConfig = {
name: 'google',
config: {
api: 'google-generative-ai',
baseUrl: 'https://generativelanguage.googleapis.com/v1beta',
apiKey: 'GEMINI_API_KEY',
},
models: [
{
modelId: 'gemini-2.0-flash',
name: 'Gemini 2.0 Flash',
config: {
api: 'google-generative-ai',
reasoning: false,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 8192,
cost: {
input: 0.1,
output: 0.4,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-2.0-flash-lite',
name: 'Gemini 2.0 Flash-Lite',
config: {
api: 'google-generative-ai',
reasoning: false,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 8192,
cost: {
input: 0.075,
output: 0.3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-2.5-flash',
name: 'Gemini 2.5 Flash',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.3,
output: 2.5,
cacheRead: 0.03,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-2.5-flash-lite',
name: 'Gemini 2.5 Flash-Lite',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.1,
output: 0.4,
cacheRead: 0.01,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-2.5-pro',
name: 'Gemini 2.5 Pro',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3-flash-preview',
name: 'Gemini 3 Flash Preview',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.5,
output: 3,
cacheRead: 0.05,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3-pro-preview',
name: 'Gemini 3 Pro Preview',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 2,
output: 12,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-flash-lite',
name: 'Gemini 3.1 Flash Lite',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.25,
output: 1.5,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-flash-lite-preview',
name: 'Gemini 3.1 Flash Lite Preview',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.25,
output: 1.5,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-pro-preview',
name: 'Gemini 3.1 Pro Preview',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 2,
output: 12,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-pro-preview-customtools',
name: 'Gemini 3.1 Pro Preview Custom Tools',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 2,
output: 12,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.5-flash',
name: 'Gemini 3.5 Flash',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.5,
output: 9,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.5-flash-lite',
name: 'Gemini 3.5 Flash Lite',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.3,
output: 2.5,
cacheRead: 0.03,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.6-flash',
name: 'Gemini 3.6 Flash',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.5,
output: 7.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-flash-latest',
name: 'Gemini Flash Latest',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.5,
output: 9,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-flash-lite-latest',
name: 'Gemini Flash-Lite Latest',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.25,
output: 1.5,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gemma-4-26b-a4b-it',
name: 'Gemma 4 26B A4B IT',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 32768,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gemma-4-31b-it',
name: 'Gemma 4 31B IT',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 32768,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,133 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'groq'
export const providerConfig = {
name: 'groq',
config: {
api: 'openai-completions',
baseUrl: 'https://api.groq.com/openai/v1',
apiKey: 'GROQ_API_KEY',
},
models: [
{
modelId: 'llama-3.1-8b-instant',
name: 'Llama 3.1 8B',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 131072,
maxTokens: 131072,
cost: {
input: 0.05,
output: 0.08,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'llama-3.3-70b-versatile',
name: 'Llama 3.3 70B',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 131072,
maxTokens: 32768,
cost: {
input: 0.59,
output: 0.79,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'meta-llama/llama-4-scout-17b-16e-instruct',
name: 'Llama 4 Scout 17B 16E',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text', 'image'],
contextWindow: 131072,
maxTokens: 8192,
cost: {
input: 0.11,
output: 0.34,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'openai/gpt-oss-120b',
name: 'GPT OSS 120B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 65536,
cost: {
input: 0.15,
output: 0.6,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
{
modelId: 'openai/gpt-oss-20b',
name: 'GPT OSS 20B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 65536,
cost: {
input: 0.075,
output: 0.3,
cacheRead: 0.0375,
cacheWrite: 0,
},
},
},
{
modelId: 'openai/gpt-oss-safeguard-20b',
name: 'Safety GPT OSS 20B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 65536,
cost: {
input: 0.075,
output: 0.3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'qwen/qwen3-32b',
name: 'Qwen3-32B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 40960,
cost: {
input: 0.29,
output: 0.59,
cacheRead: 0,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,847 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'huggingface'
export const providerConfig = {
name: 'huggingface',
config: {
api: 'openai-completions',
baseUrl: 'https://router.huggingface.co/v1',
apiKey: 'HF_TOKEN',
},
models: [
{
modelId: 'MiniMaxAI/MiniMax-M2',
name: 'MiniMax-M2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 128000,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'MiniMaxAI/MiniMax-M2.1',
name: 'MiniMax-M2.1',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'MiniMaxAI/MiniMax-M2.5',
name: 'MiniMax-M2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.03,
cacheWrite: 0,
},
},
},
{
modelId: 'MiniMaxAI/MiniMax-M2.7',
name: 'MiniMax-M2.7',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
{
modelId: 'MiniMaxAI/MiniMax-M3',
name: 'MiniMax-M3',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 524288,
maxTokens: 128000,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3-235B-A22B',
name: 'Qwen3 235B-A22B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 40960,
maxTokens: 16384,
cost: {
input: 0.2,
output: 0.8,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3-235B-A22B-Thinking-2507',
name: 'Qwen3-235B-A22B-Thinking-2507',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 131072,
cost: {
input: 0.3,
output: 3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3-32B',
name: 'Qwen3 32B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 16384,
cost: {
input: 0.29,
output: 0.59,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3-Coder-30B-A3B-Instruct',
name: 'Qwen3-Coder 30B-A3B Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.07,
output: 0.26,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3-Coder-480B-A35B-Instruct',
name: 'Qwen3-Coder-480B-A35B-Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 66536,
cost: {
input: 2,
output: 2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3-Coder-Next',
name: 'Qwen3-Coder-Next',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.2,
output: 1.5,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3-Next-80B-A3B-Instruct',
name: 'Qwen3-Next-80B-A3B-Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 66536,
cost: {
input: 0.25,
output: 1,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3-Next-80B-A3B-Thinking',
name: 'Qwen3-Next-80B-A3B-Thinking',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 131072,
cost: {
input: 0.3,
output: 2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3.5-122B-A10B',
name: 'Qwen3.5 122B-A10B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.4,
output: 3.2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3.5-27B',
name: 'Qwen3.5 27B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.3,
output: 2.4,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3.5-35B-A3B',
name: 'Qwen3.5 35B-A3B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.25,
output: 2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3.5-397B-A17B',
name: 'Qwen3.5-397B-A17B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 32768,
cost: {
input: 0.6,
output: 3.6,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3.5-9B',
name: 'Qwen3.5 9B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.17,
output: 0.25,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3.6-27B',
name: 'Qwen3.6 27B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.47,
output: 3.19,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'Qwen/Qwen3.6-35B-A3B',
name: 'Qwen3.6 35B-A3B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.15,
output: 0.95,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'XiaomiMiMo/MiMo-V2-Flash',
name: 'MiMo-V2-Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 4096,
cost: {
input: 0.1,
output: 0.3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'XiaomiMiMo/MiMo-V2.5-Pro',
name: 'MiMo-V2.5-Pro',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1048576,
maxTokens: 131072,
cost: {
input: 1,
output: 3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-ai/DeepSeek-R1',
name: 'DeepSeek-R1',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 64000,
maxTokens: 32768,
cost: {
input: 0.7,
output: 2.5,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-ai/DeepSeek-R1-0528',
name: 'DeepSeek-R1-0528',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 163840,
maxTokens: 163840,
cost: {
input: 3,
output: 5,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-ai/DeepSeek-V3.2',
name: 'DeepSeek-V3.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 163840,
maxTokens: 65536,
cost: {
input: 0.28,
output: 0.4,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-ai/DeepSeek-V4-Flash',
name: 'DeepSeek V4 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1048576,
maxTokens: 384000,
cost: {
input: 0.14,
output: 0.28,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-ai/DeepSeek-V4-Pro',
name: 'DeepSeek V4 Pro',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1048576,
maxTokens: 393216,
cost: {
input: 0.435,
output: 0.87,
cacheRead: 0.003625,
cacheWrite: 0,
},
},
},
{
modelId: 'google/gemma-4-26B-A4B-it',
name: 'Gemma 4 26B A4B IT',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 32768,
cost: {
input: 0.13,
output: 0.4,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'google/gemma-4-31B-it',
name: 'Gemma 4 31B IT',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 32768,
cost: {
input: 0.14,
output: 0.4,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'meta-llama/Llama-3.3-70B-Instruct',
name: 'Llama-3.3-70B-Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 131072,
maxTokens: 4096,
cost: {
input: 0.59,
output: 0.79,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'moonshotai/Kimi-K2-Instruct',
name: 'Kimi-K2-Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 131072,
maxTokens: 16384,
cost: {
input: 1,
output: 3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'moonshotai/Kimi-K2-Instruct-0905',
name: 'Kimi-K2-Instruct-0905',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 16384,
cost: {
input: 1,
output: 3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'moonshotai/Kimi-K2-Thinking',
name: 'Kimi-K2-Thinking',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.6,
output: 2.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'moonshotai/Kimi-K2.5',
name: 'Kimi-K2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.6,
output: 3,
cacheRead: 0.1,
cacheWrite: 0,
},
},
},
{
modelId: 'moonshotai/Kimi-K2.6',
name: 'Kimi-K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.16,
cacheWrite: 0,
},
},
},
{
modelId: 'moonshotai/Kimi-K2.7-Code',
name: 'Kimi K2.7 Code',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'openai/gpt-oss-120b',
name: 'GPT OSS 120B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 32768,
cost: {
input: 0.25,
output: 0.69,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'openai/gpt-oss-20b',
name: 'GPT OSS 20B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 32768,
cost: {
input: 0.1,
output: 0.5,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'stepfun-ai/Step-3.5-Flash',
name: 'Step 3.5 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 256000,
cost: {
input: 0.1,
output: 0.3,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'stepfun-ai/Step-3.7-Flash',
name: 'Step 3.7 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 256000,
cost: {
input: 0.2,
output: 1.15,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-4.5',
name: 'GLM-4.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 98304,
cost: {
input: 0.6,
output: 2.2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-4.5-Air',
name: 'GLM-4.5-Air',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 98304,
cost: {
input: 0.13,
output: 0.85,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-4.5V',
name: 'GLM-4.5V',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 65536,
maxTokens: 16384,
cost: {
input: 0.6,
output: 1.8,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-4.6',
name: 'GLM-4.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.55,
output: 2.2,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-4.7',
name: 'GLM-4.7',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.6,
output: 2.2,
cacheRead: 0.11,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-4.7-Flash',
name: 'GLM-4.7-Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 200000,
maxTokens: 128000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-5',
name: 'GLM-5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 202752,
maxTokens: 131072,
cost: {
input: 1,
output: 3.2,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-5.1',
name: 'GLM-5.1',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 202752,
maxTokens: 131072,
cost: {
input: 1,
output: 3.2,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'zai-org/GLM-5.2',
name: 'GLM-5.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 131072,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,65 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'kimi-coding'
export const providerConfig = {
name: 'kimi-coding',
config: {
api: 'anthropic-messages',
baseUrl: 'https://api.kimi.com/coding',
apiKey: 'KIMI_API_KEY',
},
models: [
{
modelId: 'k3',
name: 'Kimi K3',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 131072,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-for-coding',
name: 'Kimi K2.7 Code',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 32768,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.19,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-for-coding-highspeed',
name: 'Kimi For Coding HighSpeed',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 32768,
cost: {
input: 1.9,
output: 8,
cacheRead: 0.38,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,65 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'minimax-cn'
export const providerConfig = {
name: 'minimax-cn',
config: {
api: 'anthropic-messages',
baseUrl: 'https://api.minimaxi.com/anthropic',
apiKey: 'MINIMAX_CN_API_KEY',
},
models: [
{
modelId: 'MiniMax-M2.7',
name: 'MiniMax-M2.7',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0.375,
},
},
},
{
modelId: 'MiniMax-M2.7-highspeed',
name: 'MiniMax-M2.7-highspeed',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.6,
output: 2.4,
cacheRead: 0.06,
cacheWrite: 0.375,
},
},
},
{
modelId: 'MiniMax-M3',
name: 'MiniMax-M3',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,65 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'minimax'
export const providerConfig = {
name: 'minimax',
config: {
api: 'anthropic-messages',
baseUrl: 'https://api.minimax.io/anthropic',
apiKey: 'MINIMAX_API_KEY',
},
models: [
{
modelId: 'MiniMax-M2.7',
name: 'MiniMax-M2.7',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0.375,
},
},
},
{
modelId: 'MiniMax-M2.7-highspeed',
name: 'MiniMax-M2.7-highspeed',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.6,
output: 2.4,
cacheRead: 0.06,
cacheWrite: 0.375,
},
},
},
{
modelId: 'MiniMax-M3',
name: 'MiniMax-M3',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,524 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'mistral'
export const providerConfig = {
name: 'mistral',
config: {
api: 'mistral-conversations',
baseUrl: 'https://api.mistral.ai',
apiKey: 'MISTRAL_API_KEY',
},
models: [
{
modelId: 'codestral-latest',
name: 'Codestral (latest)',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 256000,
maxTokens: 4096,
cost: {
input: 0.3,
output: 0.9,
cacheRead: 0.03,
cacheWrite: 0,
},
},
},
{
modelId: 'devstral-2512',
name: 'Devstral 2',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.4,
output: 2,
cacheRead: 0.04,
cacheWrite: 0,
},
},
},
{
modelId: 'devstral-latest',
name: 'Devstral 2',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.4,
output: 2,
cacheRead: 0.04,
cacheWrite: 0,
},
},
},
{
modelId: 'devstral-medium-2507',
name: 'Devstral Medium',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.4,
output: 2,
cacheRead: 0.04,
cacheWrite: 0,
},
},
},
{
modelId: 'devstral-medium-latest',
name: 'Devstral 2 (latest)',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.4,
output: 2,
cacheRead: 0.04,
cacheWrite: 0,
},
},
},
{
modelId: 'devstral-small-2505',
name: 'Devstral Small 2505',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.1,
output: 0.3,
cacheRead: 0.01,
cacheWrite: 0,
},
},
},
{
modelId: 'devstral-small-2507',
name: 'Devstral Small',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.1,
output: 0.3,
cacheRead: 0.01,
cacheWrite: 0,
},
},
},
{
modelId: 'labs-devstral-small-2512',
name: 'Devstral Small 2',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 256000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'magistral-medium-latest',
name: 'Magistral Medium (latest)',
config: {
api: 'mistral-conversations',
reasoning: true,
input: ['text'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2,
output: 5,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'magistral-small',
name: 'Magistral Small',
config: {
api: 'mistral-conversations',
reasoning: true,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.5,
output: 1.5,
cacheRead: 0.05,
cacheWrite: 0,
},
},
},
{
modelId: 'ministral-3b-latest',
name: 'Ministral 3B (latest)',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.04,
output: 0.04,
cacheRead: 0.004,
cacheWrite: 0,
},
},
},
{
modelId: 'ministral-8b-latest',
name: 'Ministral 8B (latest)',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.1,
output: 0.1,
cacheRead: 0.01,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-large-2411',
name: 'Mistral Large 2.1',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 131072,
maxTokens: 16384,
cost: {
input: 2,
output: 6,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-large-2512',
name: 'Mistral Large 3',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.5,
output: 1.5,
cacheRead: 0.05,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-large-latest',
name: 'Mistral Large (latest)',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.5,
output: 1.5,
cacheRead: 0.05,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-medium-2505',
name: 'Mistral Medium 3',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text', 'image'],
contextWindow: 131072,
maxTokens: 131072,
cost: {
input: 0.4,
output: 2,
cacheRead: 0.04,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-medium-2508',
name: 'Mistral Medium 3.1',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.4,
output: 2,
cacheRead: 0.04,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-medium-2604',
name: 'Mistral Medium 3.5',
config: {
api: 'mistral-conversations',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.5,
output: 7.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-medium-3.5',
name: 'Mistral Medium 3.5',
config: {
api: 'mistral-conversations',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.5,
output: 7.5,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-medium-latest',
name: 'Mistral Medium (latest)',
config: {
api: 'mistral-conversations',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.5,
output: 7.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-nemo',
name: 'Mistral Nemo',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.15,
output: 0.15,
cacheRead: 0.015,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-small-2506',
name: 'Mistral Small 3.2',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 0.1,
output: 0.3,
cacheRead: 0.01,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-small-2603',
name: 'Mistral Small 4',
config: {
api: 'mistral-conversations',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 256000,
cost: {
input: 0.15,
output: 0.6,
cacheRead: 0.015,
cacheWrite: 0,
},
},
},
{
modelId: 'mistral-small-latest',
name: 'Mistral Small (latest)',
config: {
api: 'mistral-conversations',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 256000,
cost: {
input: 0.15,
output: 0.6,
cacheRead: 0.015,
cacheWrite: 0,
},
},
},
{
modelId: 'open-mistral-7b',
name: 'Mistral 7B',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 8000,
maxTokens: 8000,
cost: {
input: 0.25,
output: 0.25,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'open-mistral-nemo',
name: 'Open Mistral Nemo',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.15,
output: 0.15,
cacheRead: 0.015,
cacheWrite: 0,
},
},
},
{
modelId: 'open-mixtral-8x22b',
name: 'Mixtral 8x22B',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 64000,
maxTokens: 64000,
cost: {
input: 2,
output: 6,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'open-mixtral-8x7b',
name: 'Mixtral 8x7B',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text'],
contextWindow: 32000,
maxTokens: 32000,
cost: {
input: 0.7,
output: 0.7,
cacheRead: 0.07,
cacheWrite: 0,
},
},
},
{
modelId: 'pixtral-12b',
name: 'Pixtral 12B',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 0.15,
output: 0.15,
cacheRead: 0.015,
cacheWrite: 0,
},
},
},
{
modelId: 'pixtral-large-latest',
name: 'Pixtral Large (latest)',
config: {
api: 'mistral-conversations',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 2,
output: 6,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,184 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'moonshotai-cn'
export const providerConfig = {
name: 'moonshotai-cn',
config: {
api: 'openai-completions',
baseUrl: 'https://api.moonshot.cn/v1',
apiKey: 'MOONSHOT_API_KEY',
},
models: [
{
modelId: 'kimi-k2-0711-preview',
name: 'Kimi K2 0711',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 131072,
maxTokens: 16384,
cost: {
input: 0.6,
output: 2.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2-0905-preview',
name: 'Kimi K2 0905',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.6,
output: 2.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2-thinking',
name: 'Kimi K2 Thinking',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.6,
output: 2.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2-thinking-turbo',
name: 'Kimi K2 Thinking Turbo',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.15,
output: 8,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2-turbo-preview',
name: 'Kimi K2 Turbo',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 2.4,
output: 10,
cacheRead: 0.6,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.5',
name: 'Kimi K2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.6,
output: 3,
cacheRead: 0.1,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.6',
name: 'Kimi K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.16,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.7-code',
name: 'Kimi K2.7 Code',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.19,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.7-code-highspeed',
name: 'Kimi K2.7 Code HighSpeed',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.9,
output: 8,
cacheRead: 0.38,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k3',
name: 'Kimi K3',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 131072,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,184 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'moonshotai'
export const providerConfig = {
name: 'moonshotai',
config: {
api: 'openai-completions',
baseUrl: 'https://api.moonshot.ai/v1',
apiKey: 'MOONSHOT_API_KEY',
},
models: [
{
modelId: 'kimi-k2-0711-preview',
name: 'Kimi K2 0711',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 131072,
maxTokens: 16384,
cost: {
input: 0.6,
output: 2.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2-0905-preview',
name: 'Kimi K2 0905',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.6,
output: 2.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2-thinking',
name: 'Kimi K2 Thinking',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.6,
output: 2.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2-thinking-turbo',
name: 'Kimi K2 Thinking Turbo',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.15,
output: 8,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2-turbo-preview',
name: 'Kimi K2 Turbo',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 2.4,
output: 10,
cacheRead: 0.6,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.5',
name: 'Kimi K2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.6,
output: 3,
cacheRead: 0.1,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.6',
name: 'Kimi K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.16,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.7-code',
name: 'Kimi K2.7 Code',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.19,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.7-code-highspeed',
name: 'Kimi K2.7 Code HighSpeed',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 1.9,
output: 8,
cacheRead: 0.38,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k3',
name: 'Kimi K3',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 131072,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,337 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'nvidia'
export const providerConfig = {
name: 'nvidia',
config: {
api: 'openai-completions',
baseUrl: 'https://integrate.api.nvidia.com/v1',
apiKey: '',
},
models: [
{
modelId: 'meta/llama-3.1-70b-instruct',
name: 'Llama 3.1 70b Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 4096,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'meta/llama-3.1-8b-instruct',
name: 'Llama 3.1 8B Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 16000,
maxTokens: 4096,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'meta/llama-3.2-11b-vision-instruct',
name: 'Llama 3.2 11b Vision Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 4096,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'meta/llama-3.2-90b-vision-instruct',
name: 'Llama-3.2-90B-Vision-Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 8192,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'meta/llama-3.3-70b-instruct',
name: 'Llama 3.3 70b Instruct',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text'],
contextWindow: 128000,
maxTokens: 4096,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'minimaxai/minimax-m3',
name: 'MiniMax-M3',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 16384,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'mistralai/mistral-large-3-675b-instruct-2512',
name: 'Mistral Large 3 675B Instruct 2512',
config: {
api: 'openai-completions',
reasoning: false,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'mistralai/mistral-small-4-119b-2603',
name: 'mistral-small-4-119b-2603',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 8192,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'moonshotai/kimi-k2.6',
name: 'Kimi K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'nvidia/nemotron-3-nano-30b-a3b',
name: 'nemotron-3-nano-30b-a3b',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 131072,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'nvidia/nemotron-3-nano-omni-30b-a3b-reasoning',
name: 'Nemotron 3 Nano Omni',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 65536,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'nvidia/nemotron-3-super-120b-a12b',
name: 'Nemotron 3 Super',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.2,
output: 0.8,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'nvidia/nemotron-3-ultra-550b-a55b',
name: 'Nemotron 3 Ultra 550B A55B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 65536,
cost: {
input: 0.5,
output: 2.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'nvidia/nvidia-nemotron-nano-9b-v2',
name: 'nvidia-nemotron-nano-9b-v2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 131072,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'openai/gpt-oss-120b',
name: 'GPT-OSS-120B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 128000,
maxTokens: 8192,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'openai/gpt-oss-20b',
name: 'GPT OSS 20B',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 32768,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'stepfun-ai/step-3.5-flash',
name: 'Step 3.5 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 256000,
maxTokens: 16384,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'stepfun-ai/step-3.7-flash',
name: 'Step 3.7 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 16384,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'z-ai/glm-5.2',
name: 'GLM-5.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 131072,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,178 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'openai-codex'
export const providerConfig = {
name: 'openai-codex',
config: {
api: 'openai-codex-responses',
baseUrl: 'https://chatgpt.com/backend-api',
apiKey: '',
},
models: [
{
modelId: 'gpt-5.3-codex-spark',
name: 'GPT-5.3 Codex Spark',
config: {
api: 'openai-codex-responses',
reasoning: true,
input: ['text'],
contextWindow: 128000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4',
name: 'GPT-5.4',
config: {
api: 'openai-codex-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 272000,
input: 5,
output: 22.5,
cacheRead: 0.5,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gpt-5.4-mini',
name: 'GPT-5.4 mini',
config: {
api: 'openai-codex-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 0.75,
output: 4.5,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.5',
name: 'GPT-5.5',
config: {
api: 'openai-codex-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 272000,
input: 10,
output: 45,
cacheRead: 1,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gpt-5.6-luna',
name: 'GPT-5.6 Luna',
config: {
api: 'openai-codex-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 1,
output: 6,
cacheRead: 0.1,
cacheWrite: 1.25,
tiers: [
{
inputTokensAbove: 272000,
input: 2,
output: 9,
cacheRead: 0.2,
cacheWrite: 2.5,
},
],
},
},
},
{
modelId: 'gpt-5.6-sol',
name: 'GPT-5.6 Sol',
config: {
api: 'openai-codex-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 6.25,
tiers: [
{
inputTokensAbove: 272000,
input: 10,
output: 45,
cacheRead: 1,
cacheWrite: 12.5,
},
],
},
},
},
{
modelId: 'gpt-5.6-terra',
name: 'GPT-5.6 Terra',
config: {
api: 'openai-codex-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 3.125,
tiers: [
{
inputTokensAbove: 272000,
input: 5,
output: 22.5,
cacheRead: 0.5,
cacheWrite: 6.25,
},
],
},
},
},
],
} as const
@@ -0,0 +1,859 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'openai'
export const providerConfig = {
name: 'openai',
config: {
api: 'openai-responses',
baseUrl: 'https://api.openai.com/v1',
apiKey: 'OPENAI_API_KEY',
},
models: [
{
modelId: 'gpt-4',
name: 'GPT-4',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text'],
contextWindow: 8192,
maxTokens: 8192,
cost: {
input: 30,
output: 60,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4-turbo',
name: 'GPT-4 Turbo',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 4096,
cost: {
input: 10,
output: 30,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4.1',
name: 'GPT-4.1',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 1047576,
maxTokens: 32768,
cost: {
input: 2,
output: 8,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4.1-mini',
name: 'GPT-4.1 mini',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 1047576,
maxTokens: 32768,
cost: {
input: 0.4,
output: 1.6,
cacheRead: 0.1,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4.1-nano',
name: 'GPT-4.1 nano',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 1047576,
maxTokens: 32768,
cost: {
input: 0.1,
output: 0.4,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o',
name: 'GPT-4o',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2.5,
output: 10,
cacheRead: 1.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-2024-05-13',
name: 'GPT-4o (2024-05-13)',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 4096,
cost: {
input: 5,
output: 15,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-2024-08-06',
name: 'GPT-4o (2024-08-06)',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2.5,
output: 10,
cacheRead: 1.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-2024-11-20',
name: 'GPT-4o (2024-11-20)',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 2.5,
output: 10,
cacheRead: 1.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-4o-mini',
name: 'GPT-4o mini',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 0.15,
output: 0.6,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5',
name: 'GPT-5',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-chat-latest',
name: 'GPT-5 Chat Latest',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-codex',
name: 'GPT-5-Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-mini',
name: 'GPT-5 Mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.25,
output: 2,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-nano',
name: 'GPT-5 Nano',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.05,
output: 0.4,
cacheRead: 0.005,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-pro',
name: 'GPT-5 Pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 15,
output: 120,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1',
name: 'GPT-5.1',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-chat-latest',
name: 'GPT-5.1 Chat',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex',
name: 'GPT-5.1 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex-max',
name: 'GPT-5.1 Codex Max',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex-mini',
name: 'GPT-5.1 Codex mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.25,
output: 2,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2',
name: 'GPT-5.2',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-chat-latest',
name: 'GPT-5.2 Chat',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-codex',
name: 'GPT-5.2 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-pro',
name: 'GPT-5.2 Pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 21,
output: 168,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-chat-latest',
name: 'GPT-5.3 Chat (latest)',
config: {
api: 'openai-responses',
reasoning: false,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 16384,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-codex',
name: 'GPT-5.3 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-codex-spark',
name: 'GPT-5.3 Codex Spark',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 32000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4',
name: 'GPT-5.4',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 272000,
input: 5,
output: 22.5,
cacheRead: 0.5,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gpt-5.4-mini',
name: 'GPT-5.4 mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.75,
output: 4.5,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-nano',
name: 'GPT-5.4 nano',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.2,
output: 1.25,
cacheRead: 0.02,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-pro',
name: 'GPT-5.4 Pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 30,
output: 180,
cacheRead: 0,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 272000,
input: 60,
output: 270,
cacheRead: 0,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gpt-5.5',
name: 'GPT-5.5',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 272000,
input: 10,
output: 45,
cacheRead: 1,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gpt-5.5-pro',
name: 'GPT-5.5 Pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 30,
output: 180,
cacheRead: 0,
cacheWrite: 0,
tiers: [
{
inputTokensAbove: 272000,
input: 60,
output: 270,
cacheRead: 0,
cacheWrite: 0,
},
],
},
},
},
{
modelId: 'gpt-5.6-luna',
name: 'GPT-5.6 Luna',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 1,
output: 6,
cacheRead: 0.1,
cacheWrite: 1.25,
tiers: [
{
inputTokensAbove: 272000,
input: 2,
output: 9,
cacheRead: 0.2,
cacheWrite: 2.5,
},
],
},
},
},
{
modelId: 'gpt-5.6-sol',
name: 'GPT-5.6 Sol',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 6.25,
tiers: [
{
inputTokensAbove: 272000,
input: 10,
output: 45,
cacheRead: 1,
cacheWrite: 12.5,
},
],
},
},
},
{
modelId: 'gpt-5.6-terra',
name: 'GPT-5.6 Terra',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 3.125,
tiers: [
{
inputTokensAbove: 272000,
input: 5,
output: 22.5,
cacheRead: 0.5,
cacheWrite: 6.25,
},
],
},
},
},
{
modelId: 'gpt-realtime-2.1',
name: 'GPT-Realtime-2.1',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 128000,
maxTokens: 32000,
cost: {
input: 4,
output: 24,
cacheRead: 0.4,
cacheWrite: 0,
},
},
},
{
modelId: 'o1',
name: 'o1',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 15,
output: 60,
cacheRead: 7.5,
cacheWrite: 0,
},
},
},
{
modelId: 'o1-pro',
name: 'o1-pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 150,
output: 600,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'o3',
name: 'o3',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 2,
output: 8,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'o3-deep-research',
name: 'o3-deep-research',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 10,
output: 40,
cacheRead: 2.5,
cacheWrite: 0,
},
},
},
{
modelId: 'o3-mini',
name: 'o3-mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 1.1,
output: 4.4,
cacheRead: 0.55,
cacheWrite: 0,
},
},
},
{
modelId: 'o3-pro',
name: 'o3-pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 20,
output: 80,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'o4-mini',
name: 'o4-mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 1.1,
output: 4.4,
cacheRead: 0.275,
cacheWrite: 0,
},
},
},
{
modelId: 'o4-mini-deep-research',
name: 'o4-mini-deep-research',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 100000,
cost: {
input: 2,
output: 8,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
],
} as const
@@ -0,0 +1,269 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'opencode-go'
export const providerConfig = {
name: 'opencode-go',
config: {
api: 'openai-completions',
baseUrl: 'https://opencode.ai/zen/go/v1',
apiKey: 'OPENCODE_API_KEY',
},
models: [
{
modelId: 'deepseek-v4-flash',
name: 'DeepSeek V4 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 0.14,
output: 0.28,
cacheRead: 0.0028,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-v4-pro',
name: 'DeepSeek V4 Pro',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 0.435,
output: 0.87,
cacheRead: 0.003625,
cacheWrite: 0,
},
},
},
{
modelId: 'glm-5.1',
name: 'GLM-5.1',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 202752,
maxTokens: 32768,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0.26,
cacheWrite: 0,
},
},
},
{
modelId: 'glm-5.2',
name: 'GLM-5.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 131072,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0.26,
cacheWrite: 0,
},
},
},
{
modelId: 'grok-4.5',
name: 'Grok 4.5',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 500000,
maxTokens: 500000,
cost: {
input: 2,
output: 6,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.6',
name: 'Kimi K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.16,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.7-code',
name: 'Kimi K2.7 Code',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.19,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k3',
name: 'Kimi K3 (2x usage)',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 131072,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 0,
},
},
},
{
modelId: 'mimo-v2.5',
name: 'MiMo V2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 0.14,
output: 0.28,
cacheRead: 0.0028,
cacheWrite: 0,
},
},
},
{
modelId: 'mimo-v2.5-pro',
name: 'MiMo V2.5 Pro',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1048576,
maxTokens: 128000,
cost: {
input: 0.435,
output: 0.87,
cacheRead: 0.003625,
cacheWrite: 0,
},
},
},
{
modelId: 'minimax-m2.7',
name: 'MiniMax-M2.7',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
{
modelId: 'minimax-m3',
name: 'MiniMax-M3',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
{
modelId: 'qwen3.6-plus',
name: 'Qwen3.6 Plus',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 65536,
cost: {
input: 0.5,
output: 3,
cacheRead: 0.05,
cacheWrite: 0.625,
},
},
},
{
modelId: 'qwen3.7-max',
name: 'Qwen3.7 Max',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 65536,
cost: {
input: 2.5,
output: 7.5,
cacheRead: 0.5,
cacheWrite: 3.125,
},
},
},
{
modelId: 'qwen3.7-plus',
name: 'Qwen3.7 Plus',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 65536,
cost: {
input: 0.4,
output: 1.6,
cacheRead: 0.04,
cacheWrite: 0.5,
},
},
},
],
} as const
@@ -0,0 +1,966 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'opencode'
export const providerConfig = {
name: 'opencode',
config: {
api: 'openai-completions',
baseUrl: 'https://opencode.ai/zen/v1',
apiKey: 'OPENCODE_API_KEY',
},
models: [
{
modelId: 'big-pickle',
name: 'Big Pickle',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'claude-fable-5',
name: 'Claude Fable 5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 10,
output: 50,
cacheRead: 1,
cacheWrite: 12.5,
},
},
},
{
modelId: 'claude-haiku-4-5',
name: 'Claude Haiku 4.5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 1,
output: 5,
cacheRead: 0.1,
cacheWrite: 1.25,
},
},
},
{
modelId: 'claude-opus-4-1',
name: 'Claude Opus 4.1',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 15,
output: 75,
cacheRead: 1.5,
cacheWrite: 18.75,
},
},
},
{
modelId: 'claude-opus-4-5',
name: 'Claude Opus 4.5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-6',
name: 'Claude Opus 4.6',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-7',
name: 'Claude Opus 4.7',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-opus-4-8',
name: 'Claude Opus 4.8',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 5,
output: 25,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'claude-sonnet-4',
name: 'Claude Sonnet 4',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-4-5',
name: 'Claude Sonnet 4.5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 64000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-4-6',
name: 'Claude Sonnet 4.6',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 64000,
cost: {
input: 3,
output: 15,
cacheRead: 0.3,
cacheWrite: 3.75,
},
},
},
{
modelId: 'claude-sonnet-5',
name: 'Claude Sonnet 5',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 2,
output: 10,
cacheRead: 0.2,
cacheWrite: 2.5,
},
},
},
{
modelId: 'deepseek-v4-flash',
name: 'DeepSeek V4 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 0.14,
output: 0.28,
cacheRead: 0.028,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-v4-flash-free',
name: 'DeepSeek V4 Flash Free',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 200000,
maxTokens: 128000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-v4-pro',
name: 'DeepSeek V4 Pro',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 1.74,
output: 3.84,
cacheRead: 0.145,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3-flash',
name: 'Gemini 3 Flash',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.5,
output: 3,
cacheRead: 0.05,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.1-pro',
name: 'Gemini 3.1 Pro Preview',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 2,
output: 12,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.5-flash',
name: 'Gemini 3.5 Flash',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.5,
output: 9,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.5-flash-lite',
name: 'Gemini 3.5 Flash Lite',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 0.3,
output: 2.5,
cacheRead: 0.03,
cacheWrite: 0,
},
},
},
{
modelId: 'gemini-3.6-flash',
name: 'Gemini 3.6 Flash',
config: {
api: 'google-generative-ai',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1048576,
maxTokens: 65536,
cost: {
input: 1.5,
output: 7.5,
cacheRead: 0.15,
cacheWrite: 0,
},
},
},
{
modelId: 'glm-5',
name: 'GLM-5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 1,
output: 3.2,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'glm-5.1',
name: 'GLM-5.1',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0.26,
cacheWrite: 0,
},
},
},
{
modelId: 'glm-5.2',
name: 'GLM-5.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 131072,
cost: {
input: 1.4,
output: 4.4,
cacheRead: 0.26,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5',
name: 'GPT-5',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.07,
output: 8.5,
cacheRead: 0.107,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-codex',
name: 'GPT-5 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.07,
output: 8.5,
cacheRead: 0.107,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5-nano',
name: 'GPT-5 Nano',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.05,
output: 0.4,
cacheRead: 0.005,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1',
name: 'GPT-5.1',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.07,
output: 8.5,
cacheRead: 0.107,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex',
name: 'GPT-5.1 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.07,
output: 8.5,
cacheRead: 0.107,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex-max',
name: 'GPT-5.1 Codex Max',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.25,
output: 10,
cacheRead: 0.125,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.1-codex-mini',
name: 'GPT-5.1 Codex Mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.25,
output: 2,
cacheRead: 0.025,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2',
name: 'GPT-5.2',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.2-codex',
name: 'GPT-5.2 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.3-codex',
name: 'GPT-5.3 Codex',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 1.75,
output: 14,
cacheRead: 0.175,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4',
name: 'GPT-5.4',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 272000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-mini',
name: 'GPT-5.4 Mini',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.75,
output: 4.5,
cacheRead: 0.075,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-nano',
name: 'GPT-5.4 Nano',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 400000,
maxTokens: 128000,
cost: {
input: 0.2,
output: 1.25,
cacheRead: 0.02,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.4-pro',
name: 'GPT-5.4 Pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 30,
output: 180,
cacheRead: 30,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.5',
name: 'GPT-5.5',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.5-pro',
name: 'GPT-5.5 Pro',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 30,
output: 180,
cacheRead: 30,
cacheWrite: 0,
},
},
},
{
modelId: 'gpt-5.6-luna',
name: 'GPT-5.6 Luna',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 1,
output: 6,
cacheRead: 0.1,
cacheWrite: 1.25,
},
},
},
{
modelId: 'gpt-5.6-sol',
name: 'GPT-5.6 Sol',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 5,
output: 30,
cacheRead: 0.5,
cacheWrite: 6.25,
},
},
},
{
modelId: 'gpt-5.6-terra',
name: 'GPT-5.6 Terra',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1050000,
maxTokens: 128000,
cost: {
input: 2.5,
output: 15,
cacheRead: 0.25,
cacheWrite: 3.125,
},
},
},
{
modelId: 'grok-4.5',
name: 'Grok 4.5',
config: {
api: 'openai-responses',
reasoning: true,
input: ['text', 'image'],
contextWindow: 500000,
maxTokens: 500000,
cost: {
input: 2,
output: 6,
cacheRead: 0.5,
cacheWrite: 0,
},
},
},
{
modelId: 'grok-build-0.1',
name: 'Grok Build 0.1',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 256000,
maxTokens: 256000,
cost: {
input: 1,
output: 2,
cacheRead: 0.2,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.5',
name: 'Kimi K2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.6,
output: 3,
cacheRead: 0.08,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.6',
name: 'Kimi K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.16,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.7-code',
name: 'Kimi K2.7 Code',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0.95,
output: 4,
cacheRead: 0.19,
cacheWrite: 0,
},
},
},
{
modelId: 'laguna-s-2.1-free',
name: 'Laguna S 2.1 Free',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 256000,
maxTokens: 32000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'mimo-v2.5-free',
name: 'MiMo V2.5 Free',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 200000,
maxTokens: 32000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'minimax-m2.5',
name: 'MiniMax-M2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
{
modelId: 'minimax-m2.7',
name: 'MiniMax-M2.7',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 204800,
maxTokens: 131072,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
{
modelId: 'minimax-m3',
name: 'MiniMax-M3',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 512000,
maxTokens: 128000,
cost: {
input: 0.3,
output: 1.2,
cacheRead: 0.06,
cacheWrite: 0,
},
},
},
{
modelId: 'nemotron-3-ultra-free',
name: 'Nemotron 3 Ultra Free',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 128000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'north-mini-code-free',
name: 'North Mini Code Free',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 256000,
maxTokens: 64000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'qwen3.5-plus',
name: 'Qwen3.5 Plus',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.2,
output: 1.2,
cacheRead: 0.02,
cacheWrite: 0.25,
},
},
},
{
modelId: 'qwen3.6-plus',
name: 'Qwen3.6 Plus',
config: {
api: 'anthropic-messages',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 65536,
cost: {
input: 0.5,
output: 3,
cacheRead: 0.05,
cacheWrite: 0.625,
},
},
},
],
} as const
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,269 @@
// This file is auto-generated by scripts/generate-builtin-providers.ts
// Do not edit manually - run 'bun run generate-providers' to update
export const providerName = 'qwen-token-plan-cn'
export const providerConfig = {
name: 'qwen-token-plan-cn',
config: {
api: 'openai-completions',
baseUrl: 'https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1',
apiKey: '',
},
models: [
{
modelId: 'MiniMax-M2.5',
name: 'MiniMax-M2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 196608,
maxTokens: 24576,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-v3.2',
name: 'DeepSeek V3.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 131072,
maxTokens: 65536,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-v4-flash',
name: 'DeepSeek V4 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'deepseek-v4-pro',
name: 'DeepSeek V4 Pro',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 384000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'glm-5',
name: 'GLM-5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 202752,
maxTokens: 16384,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'glm-5.1',
name: 'GLM-5.1',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 202752,
maxTokens: 128000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'glm-5.2',
name: 'GLM-5.2',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 131072,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.5',
name: 'Kimi K2.5',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 32768,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.6',
name: 'Kimi K2.6',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 16384,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'kimi-k2.7-code',
name: 'Kimi K2.7 Code',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 262144,
maxTokens: 262144,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'qwen3.6-flash',
name: 'Qwen3.6 Flash',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 65536,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'qwen3.6-plus',
name: 'Qwen3.6 Plus',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 65536,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'qwen3.7-max',
name: 'Qwen3.7 Max',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text'],
contextWindow: 1000000,
maxTokens: 65536,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'qwen3.7-plus',
name: 'Qwen3.7 Plus',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 64000,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
{
modelId: 'qwen3.8-max-preview',
name: 'Qwen3.8 Max Preview',
config: {
api: 'openai-completions',
reasoning: true,
input: ['text', 'image'],
contextWindow: 1000000,
maxTokens: 131072,
cost: {
input: 0,
output: 0,
cacheRead: 0,
cacheWrite: 0,
},
},
},
],
} as const

Some files were not shown because too many files have changed in this diff Show More