* feat(cli): add standalone downloads and installers
* fix(cli): silence PowerShell download progress
* fix(cli): use framework APIs for Windows verification
* fix(cli): preserve null backup path during Windows updates
* docs(cli): translate native installation quickstart into Chinese
* refactor(agent)!: migrate sessions to Pi Durable
Replace patched AgentHarness storage with per-session SQLite and PostgreSQL commit mirroring. Migrate legacy history automatically and run agent queues with the main server.
BREAKING CHANGE: Remove standalone agent workers and return native Pi entry records from chat history APIs.
* fix(agent): constrain session log dialog scrolling
Keep the header visible and wrap long output while log entries and expanded records scroll within the dialog. Cover desktop and narrow viewports through the dashboard Sessions flow.
* fix(agent): hide entry IDs from session log card headers
Retain internal entry IDs in expanded raw records for debugging. Verify that card headers omit the IDs while raw JSON preserves them.
* test(e2e): handle browser wheel limits in session log checks
Verify that a normal wheel event moves the viewport before explicitly scrolling to the final record. Keep the dialog bounds, fixed header, wrapping, and expanded-record assertions.
* fix(agent): select model providers without inheriting unrelated auth
* fix(webui): prevent resize feedback when toggling agent panel
* test(agent): reproduce concurrent comment session races
* fix(agent): serialize comment sessions and deduplicate history
* docs(changelog): consolidate Pi Durable changes into one entry
* fix: gracefully handle remote HTTP URLs for MCP and cache production static assets
* test(mcp): add remote http mcp integration test and fix static etag normalization and fallback error
* feat(storage): add team auto-archive assets policy and background job
* fix(storage): address code review feedback on lock safety, starvation, and permissions
* docs: update readme and expand archive storage documentation with auto-archive details
* feat(storage): log info message on auto-archive job start and finish with archived count
* style(storage): stack age threshold and grace period vertically with updated grace period description
* fix(storage): only log auto-archive info messages when candidate assets are found to process
* feat(storage): add archive storage support for assets
* fix(webui): show archive icon in front of file name in breadcrumb bar
* fix(webui): display archive icon in breadcrumbs when restoring
* refactor(transcode): simplify restore asset workflow to single while loop
* docs(configuration): add archive storage documentation and readme entries
* fix(webui): show archive icon in front of version names and prevent version card overflow
* fix(webui): fix overflow in manage versions dialog scroll area
* fix(webui): prevent manage versions dialog overflow on long file names
* docs(configuration): fix archive storage environment variables and defaults
* fix(storage): address code review findings for archive storage
* fix(core): strip trailing slashes when normalizing auth and MCP resource URLs
* docs: document BETTER_AUTH_URL in README and quickstart guides
* docs: prioritize AWS_ENDPOINT_URL_S3 and clarify applicability to local storage
* docs: refine AWS_ENDPOINT_URL_S3 guidance for local storage deployments
* feat(operations): move MCP to /mcp and add search, comment, metadata, version, and upload operations
- Serve the MCP server at /mcp instead of /api/mcp
- Add search_assets, copy_assets, get_folder_instructions, list_comments,
create_comment, resolve_comment, list_metadata_fields,
update_asset_metadata, create_version_stack, list_versions, and
remove_version on the agent, MCP, and CLI
- Add create_file and create_version (agent, MCP) with Creation Context
metadata, and complete_upload (MCP) for presigned two-step uploads
- Replace the native agent create_file/create_version tools with
Operation-backed ones
- Route the comment-complete, metadata, copy, version stack, and
AGENTS.md read REST endpoints through their Operations
* docs(readme): add built-in MCP server and CLI to features list
* feat(asset-read): expose read_asset on MCP and the CLI
Enable the mcp and cli surfaces for read_asset and get_asset_render.
- MCP: add per-Operation tool extensions. The read presenter attaches up
to 8 images (2 MB each) inline, lists image links, supports
inlineImages: false, and tells the model to call get_asset_render when
a render is pending. Storage keys are stripped from the result.
- CLI: generated commands can flatten object inputs into prefixed flags
(cli.flatten) and omit fields (cli.omit); add per-command extensions.
New `read` and `render` commands print image links, download them with
--out, and support --no-wait with a printed `render <taskId>` hint.
- Round frame timestamps in labels to 3 decimals.
* fix(mcp): make tool input schemas portable to single-type dialects
Nullable inputs were emitted as `type: ["number", "null"]` (or a `null`
anyOf branch), which clients that map tool schemas onto an OpenAPI-style
subset (e.g. Gemini function declarations) may reject.
- Make read_asset's startPage, endPage, and degree optional instead of
nullable.
- Move null handling into a shared dropNullOptionals helper in
@shumai/operations that also strips nulls inside nested config
objects. The agent uses it, and the MCP server applies it before the
SDK validates arguments, so models that send null keep working.
- Test that no Operation input schema (and no listed MCP tool) uses a
nullable type.
* feat(asset-download): share download_asset across agent, MCP and CLI
Turn the agent-only download_asset tool into an Operation on all three
surfaces.
- download_asset takes up to 50 file or version stack IDs and returns
presigned links. proxy (default true) picks the web preview when there
is one, as the agent always did; proxy: false returns the original.
Folders are rejected. Read is checked on every ID before anything runs.
- Agent: an extension saves the files into the .pi sandbox and keeps the
agent-only storage key mode (e.g. frames from read_asset). Agent tool
extensions gain description, parameterOverrides and intercept.
- CLI: `shumai-cli download <assetId...> [--out <dir>] [--original]`
streams files to disk without overwriting existing names.
- read_asset's unsupported-type error now names the right way to
download on each surface (download_asset, or `shumai-cli download`).
- Add the Download Asset toggle to agent settings.
Replace the monolithic agent read_asset tool with a reader registry in
@shumai/core/src/asset-read (model3d, image, video, document) behind a
shared pipeline that handles access checks, version stacks, annotations,
render tasks, and waiting.
- Add read_asset and get_asset_render Operations (agent surface only).
Renders that outlast the caller's wait budget return "pending" with the
workflow task ID; the read request is stored on the task payload so
get_asset_render works across restarts and instances (requester only).
- Add workflowService.waitFor and presign expiry support.
- Keep agent output compatible via an agent tool extension (s3KeyOnly,
image blocks, session image details); autofill keeps its team-scoped tool.
- Cap video frames at 20 per call; add the get_asset_render agent toggle.
Add a stateless Streamable HTTP MCP server at /api/mcp that exposes every
mcp-surface Operation with its input/output schemas and annotations, and
POST /api/ops/:name for the CLI. Both require an API token (session
cookies are rejected), run Operations as the token's user, honor demo
read-only mode, and record via mcp/cli/api in audit logs.
Also make list_projects toggleable in agent settings and document MCP
client setup.
* feat(workflow): migrate workflow engine from temporal to dbos
Replace Temporal and LocalExecutor with DBOS Transact (@dbos-inc/dbos-sdk) as the unified workflow engine. Background tasks, media transcoding, and AI agent workflows now execute with durable steps and task queues directly against PostgreSQL, eliminating the need for an external Temporal cluster and separate polling executors.
* fix(workflow): submit tasks via DBOSClient to allow enqueueing from steps
Enqueuing workflows using DBOS.enqueueWorkflowWithOptions threw a
DBOSInvalidWorkflowTransitionError when triggered from within a running
step or activity (e.g. creating embedding or autofill tasks during
video transcoding).
Use DBOSClient to submit and cancel workflow tasks directly against the
system database, detaching execution from the ambient workflow/step
context.
* feat(web): support SHUMAI_WORKERS environment variable
* fix(workflow): ensure CLI tasks are enqueued to DBOS and run DBOS migrations in global setup
- Initialize workflows and wait for pending DBOS submissions in shumai regenerate-embeddings CLI
- Remove no-op submit bypass and enqueue tasks via DBOSClient in producer and CLI contexts
- Add DBOS migration runner in test global setup to avoid concurrent migration deadlocks across test forks
- Update changelog and add regression tests
* fix(workflow): ensure unique per-instance executor identity for worker replicas
- Derive unique per-instance DBOS executor IDs across worker replicas
- Support SHUMAI_WORKER_ID, DBOS__VMID, and POD_NAME (Kubernetes Downward API)
- Disambiguate multi-replica bare metal/container deployments with host, PID, and instance suffix
- Prevent startup recovery from re-enqueueing workflows actively executed by other live workers
- Add multi-instance startup recovery tests and update deployment documentation
* fix(workflow): only register listened queues and update concurrency on change
- Only register queues that the executor instance actively listens to
- Use always_update conflict resolution so environment variable changes take effect on restart
- Prevent dedicated workers from overwriting unmanaged queue concurrency configurations
- Add tests verifying concurrency updates and queue isolation across worker roles
* refactor(workflow): simplify activities to use defineStep and direct calls (#482)
* refactor(workflow): simplify activities to use defineStep and direct calls
- Wrap activities using defineStep from @shumai/workflow-core
- Remove obsolete getActivities, registerActivities, dynamic activity proxies, and workerQueue arguments
- Remove executor.ts and registry.ts, consolidating workflow registration into workflow.ts
- Call activities directly across transcode and agent workflows
- Update all corresponding workflow tests and developer guide documentation
* fix(workflow): properly detect DBOSWorkflowCancelledError in runOnCancel fallback
- DBOSWorkflowCancelledError in DBOS SDK does not set name, inheriting 'Error'
- Implement isDbosWorkflowCancelledError checking instanceof, constructor.name, and dbosErrorCode
- Update unit tests to verify real DBOS SDK cancellation error instances
* docs(workflow): update documentation and readme for dbos migration
- Rename docs/configuration/temporal.mdx to docs/configuration/distributed.mdx
- Document DBOS-based distributed worker architecture and task queues
- Remove obsolete docker-compose/temporal/
- Update environment variables documentation in docs/configuration/env-variables.mdx
- Update references in README.md, README_CN.md, docs/docs.json, and related guides
* feat(deploy): implement KEDA ScaledJob and scale-to-zero On-Demand transcoding (#483)
* feat(deploy): add shumai-transcode docker image and k8s autoscaling example with garage s3
* docs(k8s): document port-forwarding and in-cluster s3 domain mapping
* feat(storage): support public s3 endpoint for browser uploads and update k8s example
* fix(k8s): configure s3 bucket cors rules in garage initialization
* fix(k8s): increase transcode worker memory limit to 4Gi for 4K video encoding
* feat(deploy): add AWS EKS deployment with Graviton3 Spot nodegroups and autoscaling
* docs(deploy): redesign EKS documentation, rename k8s to eks, and add scalable deployment feature
* docs: rephrase DBOS feature bullet to Durable Workflows via DBOS
* docs(eks): upgrade default instances to Graviton5 c9g.large and c9g.xlarge
* feat(eks): use On-Demand for general nodegroup and add spot capacity diversification
* docs(eks): fix Architecture diagram to locate all transcode worker pods in Transcode Compute Pool
* feat(deploy): implement KEDA ScaledJob and scale-to-zero On-Demand transcoding
* feat(transcode): support idle-timeout auto-drain for KEDA scaled jobs
* fix(deploy): fix EKS deployment manifests and documentation based on cluster verification
* feat(embedding): update compose tiers, docs, and add regenerate-embeddings CLI
- Add Lite, Embedding, and Full Docker Compose configurations (and Aliyun mirrors)
- Set CONCURRENCY_AGENT to 1 for compose files containing embedding-server
- Update README and README_CN with compose options, recommended RAM, and semantic search links
- Add LiteRT-LM Embedding Server setup guide in docs/using-shumai/embedding-server.mdx
- Add shumai regenerate-embeddings CLI command and EmbeddingService to regenerate embeddings across teams
- Add comprehensive test coverage for EmbeddingService
* fix(docker): remove basic auth for turntable-renderer in full compose
* fix(agent): lock and disable embedding agent when ENABLE_EMBEDDING_AGENT is false
Ensure that when ENABLE_EMBEDDING_AGENT is set to 'false', existing embedding agents are automatically disabled on startup, marked as read-only, and blocked from manual API modifications or creations. Additionally guard background task creation and semantic search when the environment variable disables embedding agents.
* test(upload): remove flaky status assertion on workflow tasks
* feat(transcode): decode and scale on the GPU for VA-API transcodes
When the selected encoder is h264_vaapi, decode with -hwaccel vaapi and
scale with scale_vaapi so frames stay in GPU memory, instead of decoding
and scaling on the CPU and hwupload-ing before the encoder. On a Haswell
iGPU this makes 4K proxy/HLS renditions 3-5x faster.
Fallback order: GPU decode + VA-API encode -> software decode + VA-API
encode (previous behavior) -> libx264. Watermark overlays, HDR sources
and rotated sources skip GPU decode: ffmpeg cannot auto-insert its
rotation filter on hardware frames, so rotated input would otherwise come
out sideways instead of failing. SHUMAI_VAAPI_HW_DECODE=false disables it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(transcode): document VA-API GPU decoding and SHUMAI_VAAPI_HW_DECODE
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* refactor(transcode): generic hardware decode config, stall watchdog, reuse probed rotation
Follow-ups from comparing with Immich and Jellyfin:
- Replace the VA-API-specific helpers with a per-encoder HW_DECODE_CONFIGS
table (input args, output args, GPU scale filter) and build the MP4 and
HLS input/filter graph in one shared method so the paths cannot drift.
Only h264_vaapi has an entry; other backends can be added once tested on
real hardware. The opt-out is now SHUMAI_HW_DECODE=false.
- Kill a GPU-decode attempt with SIGKILL when its output stops growing for
SHUMAI_HW_DECODE_STALL_TIMEOUT (default 120s) and fall back, since some
VA-API driver failures hang ffmpeg instead of exiting (immich#31933).
- Add -noautoscale on the GPU-decode path so a mid-stream decoder reinit
(e.g. 1080 -> 1088) cannot auto-insert a software scaler into the GPU
graph (jellyfin#17133).
- Pass the rotation already probed into MediaMetadata through the workflow
and activities as sourceRotation; ffprobe only runs when it is missing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* feat(settings): redesign video transcoding settings into dedicated cards and rename team AI settings
* feat(settings): rename Video Strategy to MP4 Transcoding with sub-description
* docs(settings): add email notification documentation to team settings
* fix(notification): use correct webui route URLs for project and asset links in emails
* feat(transcode): add HDR output proxy support
* fix(transcode): auto-detect MoltenVK Vulkan ICD on macOS for libplacebo filters
* fix(transcode): split filter in generateSprite to ensure poster uses tone-mapped stream
* docs(changelog): update docs/changelog.mdx and ignore in prettier
* fix(transcode): respect video rotation metadata to preserve display aspect ratio
* fix(webui): render native video with hardware EDR acceleration instead of 2D canvas
* fix(webui): show HDR label on mobile video control bar
* fix(share): preserve hdr flag on watermarked video transcodes
* fix(webui): differentiate HDR download options in video player
* fix(transcode): handle Dolby Vision Profile 5 via standard HDR pipeline without libplacebo
* fix(transcode): prepend setparams filter to prevent zscale colorspace path error for DV P5
* refactor(transcode): simplify HDR proxies to always produce HDR without SDR setting
* feat(agent): consolidate media tools into read_asset and add S3 key support to download_asset
* fix(agent): refine parameter descriptions and remove download URL from read_asset text content
* fix(agent): resolve owning asset and enforce RBAC permissions when downloading by S3 key
* feat(auth): add admin password reset CLI command and reset password web UI
* fix(auth): invalidate previous reset tokens and use AsyncLocalStorage for link capture