* feat(webui): redesign select and multi-select metadata fields with search box and dynamic option creation
* fix(webui): stabilize select popover placement and preserve field visibility in store
* refactor(webui): enforce 2-row collapsed height and whitespace-pre-wrap on LongTextField
* feat(agent): type-safe CREATION_CONTEXT metadata on create_file/create_version
Build the create_file and create_version tool metadata parameter from the
project's CREATION_CONTEXT fields at harness creation time, instead of the
schemaless key-value map that required the agent to call list_autofill_fields
first.
- Thread projectId from the chat workflow into createAgentSession and build
a strict, all-required (nullable) metadata schema via fieldsToTypeBoxSchema
(additionalProperties: false, option IDs enforced via enum).
- Remove the list_autofill_fields tool and its activity case.
- Reject metadata keys that are not CREATION_CONTEXT fields of the target
project with a clear non-retryable error telling the agent to omit metadata.
- Skip null metadata values when persisting (create_file/create_version and
the autofill workflow).
- Switch the touched schemas from @sinclair/typebox 0.34 to typebox v1 to
match the harness validator.
* refactor(agent): migrate remaining tools from @sinclair/typebox 0.34 to typebox v1
All agent tool schemas now use the same typebox v1 package as the harness
validator; drop the @sinclair/typebox 0.34 dependency from @shumai/agent.
* test(agent): reproduce null coercion in harness tool validation
The pi-ai harness (<=0.82.1) coerces null values in [string, null] unions to
'' during tool argument validation for typebox v1 schemas. A valid create_file
call with path: null + data: {...} therefore arrives at execute with path: ''
and fails with 'Provide exactly one of path or data'. Null metadata values
(the 'unknown' sentinel) are likewise turned into ''.
This test reproduces both failures. It is expected to fail until
@earendil-works/pi-ai is upgraded to >=0.84.0 (#7328), which preserves values
that already match a union arm.
* fix(agent): patch pi-ai 0.82.1 to preserve null in nullable union tool args
Apply the upstream fix for earendil-works/pi#7328 (PR #7373) via a bun
patch instead of upgrading pi-ai. coerceWithUnionSchema now returns the
original value when it already validates against any union member,
instead of coercing null in [string, null] unions into "".
Fixes the create_file false "Provide exactly one of path or data" error
when the model passes path: null with data. Reproduced by
packages/agent/src/reproduce-tool-null-coercion.test.ts, which now passes.
* fix(docker): copy bun patch dir before frozen-lockfile install
The pi-ai patchedDependencies entry makes bun install --frozen-lockfile
fail in the builder stage because patches/ was not present in the image
yet. Copy it alongside package.json/bun.lock so the patch applies.
* refactor(agent): simplify metadata guidance in system prompt and error message
Shorten the create_file/create_version metadata instructions in the system
prompt and reword the invalid-metadata-key error to not reference
CREATION_CONTEXT internals. Update the matching test assertion.
* feat(metadata): replace aiAutofill boolean with autofillSource enum
* fix(agent): validate creation-context metadata and make writes transactional
Address code review feedback on the autofillSource migration:
- Update the agent system prompt to reference the 'metadata' parameter of
create_file/create_version instead of the removed 'context' parameter,
which was being silently discarded by the activity
- Write metadata inside the caller's transaction via the new
metadataService.updateAssetMetadataInTx, keeping metadata rows atomic
with asset creation instead of opening a separate non-atomic transaction
- Reject metadata keys that are not defined CREATION_CONTEXT fields of the
project with a non-retryable ApplicationFailure before creating the asset
- Remove dead agentContext handling from autofillAiActivity and its tests,
and drop the unused ai_autofill i18n keys (paraglide recompiled)
- Derive the DTO autofillSource enum from the AutofillSource const instead
of a hardcoded z.enum to prevent drift
* docs(mcp): add MCP support documentation
* docs(readme): simplify MCP feature description
* docs(mcp): fix inaccuracies in MCP servers documentation
- Clarify that discovery runs on refresh, OAuth completion, or first
tool call, not automatically at server creation
- Note that advanced settings (transport, auth, permission, direct
tools) are configured in the edit dialog after creation
- Correct agent assignment UI description: switch toggle, not checkbox
* docs(mcp): explain proxy vs direct tool configuration
Add a dedicated section covering the three tool states (Off / Proxy /
Direct), what the mcp proxy tool does and why it exists, and when and
why to use direct tools, including the context-token trade-off.
* feat(webui): render pdf pages via live canvas with selectable text
Replace the rasterize-to-PNG pipeline in the PDF viewer with direct
canvas rendering that tracks the current zoom, so text stays crisp when
zoomed in, and overlay a pdf.js TextLayer that makes text natively
selectable when not in draw mode.
- pdf-viewer: render pages onto a live canvas at a zoom-proportional
scale (debounced, memory-capped) instead of a fixed 2x PNG snapshot
- pdf-viewer: add pdf.js TextLayer overlay per page, transformed in
lockstep with the Konva stage so spans stay aligned while panning
and zooming
- drawing-canvas: accept an optional live canvas element and force a
layer redraw when it is swapped
- add text-layer.css with the minimal pdf.js text layer styles
- add e2e spec verifying text selection in view mode and that draw
mode routes drags to the canvas
* fix(webui): show pdf text selection highlight above the page canvas
The pdf.js text layer was stacked below the opaque Konva page canvas, so
the native selection highlight was painted underneath it and invisible
(selection itself worked). Move the overlay above the canvas and gate its
pointer-events on draw mode: auto when not drawing (select text), none in
draw mode so drags reach the Konva stage below (draw markup).
Verified with screenshot pixel analysis: the selected region now renders
the highlight (blue) where it was pure grayscale before.
* fix(webui): transparent selection background for pdf text layer br elements
* fix(webui): stop pdf text selection over-selecting in empty space
The text layer container was user-selectable (user-select: auto), so when
a drag's caret moved into empty space Chromium targeted the container and
extended the selection to a container-content boundary, selecting across
rows (e.g. dragging from line 1 into the gap selected lines 1-2).
Make the container user-select: none while keeping the text spans
user-select: text, so the caret clamps to the nearest selectable text and
the selection stays put when dragging into empty space.
* fix(webui): deselect pdf text when clicking empty space
With user-select: none on the text layer container, Chromium no longer
clears the selection when clicking non-selectable areas. Clear it manually
on mousedown over the container or <br> (not a text span) so clicking
empty space between lines or outside text deselects, matching native
behavior.
* fix(webui): add content-only pinch-zoom and two-finger pan to media viewers
Intercept trackpad pinch (ctrl+wheel), touchscreen pinch (two touches),
and two-finger move inside the image, video, and PDF viewers so they
transform only the media content instead of triggering browser page
zoom (matching Frame.io behavior).
- New shared pan-zoom math (zoomAtPoint, fitScale, isZoomed) and a
usePanZoomGestures hook using native non-passive wheel/touch listeners
with touch-action: none on the media container.
- Plain wheel / two-finger scroll pans the content, but only once the
user has zoomed in; drag-pan is gated the same way.
- Image and PDF viewers no longer reset zoom to fit on container resizes
while the user has manually zoomed (matches the video viewer).
- DrawingCanvas ignores multi-touch moves so single-pointer pan/draw
doesn't fight the pinch handler; the video viewer keeps play/pause on
the container div to avoid double-toggling when the overlay is active.
Includes unit tests for the pan-zoom math.
* fix(webui): disable click-drag pan for pdf and video viewers
When zoomed, PDF and video previously panned on left-button (or
single-finger touch) drag via the shared DrawingCanvas. That conflicts
with left-click semantics: PDF wants the drag for text selection and
video reserves the click for play/pause.
Remove the gated onPan from both viewers so the Konva overlay stays
pointer-events:none when not drawing - clicks pass straight through
(video click-to-play now works cleanly at any zoom) - while panning
remains available via mouse wheel and two-finger gestures. Image keeps
drag-pan.
* feat(mcp): support tool-level directTools array and 3-state UI configuration
* refactor(mcp): refine tool hint explanation and separate server instructions block
* style(mcp): display tool count in header title and split option hints into 3 lines
* style(mcp): update tools description paragraph and place bulk buttons on a separate row
* fix(mcp): enforce excludeTools in proxy path and align direct-tool counts
- Filter the per-agent tool registry by excludeTools at build time (DB cache
and lazy-connect paths), so excluded tools are unreachable through the mcp
proxy (call, search, list, describe) — matching pi-mcp-adapter where
isToolAllowed gates both direct and gateway exposure.
- Add shared isDirectTool predicate (matches original or prefixed name) used
by both buildDirectTools and the proxy description counts, so the
advertised direct/proxy counts can no longer drift from actual
registration; toolCount now reflects allowed tools only.
- Remove the legacy boolean directTools branch (no legacy data exists).
- Drop the now-unused mcp_excluded_tools_hint i18n key and regenerate
Paraglide messages.
- Extend unit tests: excluded tools are not counted, listed, searched, or
callable via the proxy; prefixed-name directTools configs count correctly.
* fix(agent): resolve version stack name and size in list_assets
* fix(agent): resolve version stack to latest version in download_asset, analyze_image, and read_pdf_pages
* feat(mcp): add MCP server support with proxy tool and OAuth flow
Add URL-based MCP (Model Context Protocol) server support to the backend:
- DB: McpServer / AgentMcpServer / McpServerCredential models with
typed JSON columns (auth config, cached tools, stored OAuth tokens,
pending auth state) and a migration.
- Core (@shumai/core/src/mcp): DB-backed credential store, OAuth
client provider (authorization_code + client_credentials) with a
Hono-hosted redirect callback, connection lifecycle manager
(Streamable HTTP with SSE fallback, lazy connect, 401 -> needs_auth),
JSON Schema -> AgentTool parameter conversion, in-memory tool registry
with ranked search, the single 'mcp' proxy AgentTool (search /
describe / list / call / connect / status / auth-start), per-server
opt-in direct tools ({server}_{tool}), and McpService (CRUD,
discovery, OAuth, per-agent assignment via AgentMcpServer, and
permission gating on tool execution mirroring skills).
- Agent: chat agents receive the 'mcp' proxy tool (plus direct tools
for opted-in servers) built from their assigned, enabled servers.
- API: /teams/:teamId/mcp/servers CRUD, tools refresh/list, auth
start/complete/status/remove, test, permission, and the public
OAuth callback route; Admin authz via new ResourceType.McpServer;
audit logging.
- DTOs: zod schemas + types for servers, tools, auth, and per-agent
mcpServerIds.
- Tests: DB store, schema conversion, registry/search, service
integration against a real in-process MCP server (v2
@modelcontextprotocol/server), permission gating, rename
invalidation, direct tools, per-agent assignment, API route tests,
and workflow E2E (local + temporal) for proxy and direct modes.
* fix(mcp): persist OAuth discovery snapshot and stamp SEP-2352 issuer
The callback leg of the OAuth flow failed with
AuthorizationServerMismatchError ('discoveryState was not available on
the callback leg') because McpDbStore.savePendingAuth overwrote the
entire pendingAuth JSON column, clobbering the discoverySnapshot that
the provider's saveDiscoveryState() had stored inside it.
- McpDbStore.savePendingAuth now preserves an in-flight discovery
snapshot so the callback leg can satisfy the SEP-2352 authorization-
server issuer check.
- McpOauthProvider now round-trips the issuer stamp derived from the
discovery snapshot: saveClientInformation/saveTokens persist it,
clientInformation()/tokens() return it, and tokens() backfills a
missing stamp for pre-existing credentials (removes the
'stored OAuth credential has no issuer stamp' warning).
- Regression tests: savePendingAuth keeps the snapshot; the provider
stamps/round-trips the issuer on both legs.
* fix(mcp): backfill SEP-2352 issuer stamp on stale OAuth client info
clientInformation() returned stored client info without an issuer field
when it had been dynamically registered before the stamping fix, so the
SDK logged 'stored OAuth credential has no issuer stamp' on every read.
Now the stored branch backfills the issuer from the discovery snapshot
(persisting it) exactly like the tokens() backfill.
* feat(webui): add MCP server configuration and agent assignment UI (#290)
* feat(webui): add MCP server configuration and agent assignment UI
* fix(webui): auto-detect auth for unconfigured MCP servers and show connect button
* fix(mcp): auto-generate state in mcp-oauth-provider when missing in DB
* fix(mcp): persist OAuth discovery snapshot and stamp SEP-2352 issuer
The callback leg of the OAuth flow failed with
AuthorizationServerMismatchError ('discoveryState was not available on
the callback leg') because McpDbStore.savePendingAuth overwrote the
entire pendingAuth JSON column, clobbering the discoverySnapshot that
the provider's saveDiscoveryState() had stored inside it.
- McpDbStore.savePendingAuth now preserves an in-flight discovery
snapshot so the callback leg can satisfy the SEP-2352 authorization-
server issuer check.
- McpOauthProvider now round-trips the issuer stamp derived from the
discovery snapshot: saveClientInformation/saveTokens persist it,
clientInformation()/tokens() return it, and tokens() backfills a
missing stamp for pre-existing credentials (removes the
'stored OAuth credential has no issuer stamp' warning).
- Regression tests: savePendingAuth keeps the snapshot; the provider
stamps/round-trips the issuer on both legs.
* fix(mcp): backfill SEP-2352 issuer stamp on stale OAuth client info
clientInformation() returned stored client info without an issuer field
when it had been dynamically registered before the stamping fix, so the
SDK logged 'stored OAuth credential has no issuer stamp' on every read.
Now the stored branch backfills the issuer from the discovery snapshot
(persisting it) exactly like the tokens() backfill.
* fix(mcp): attach McpOauthProvider for auto-detected OAuth servers and discover tools on auth completion
* fix(webui): fix flex min-h-0 and max-h scrolling on MCP tools dialog
* fix(webui): set explicit flex height on McpToolsDialog for active scrolling
* feat(webui): make MCP server cards clickable to edit and widen server config dialog
* feat(mcp): agent-level server enablement and auto-detected server metadata (#291)
* feat(mcp): agent-level server enablement and auto-detected server metadata
- Remove the server-level `enabled` column from mcp_servers; enable/disable
now happens per-agent via the AgentMcpServer join table (mirrors skills).
- Drop the unique (teamId, name) constraint so duplicate server names are allowed.
- Add a `description` column, auto-filled from client.getServerVersion() on
connect; the server name is auto-filled from the reported title (fallback name)
with a URL-derived placeholder until first connect.
- Remove manual name input: the create dialog is URL-only, and the edit dialog
shows the auto-detected name/description read-only.
- Replace the include/exclude raw-text inputs with the tool list in the edit
dialog, managing only excludeTools; remove includeTools support and the
separate McpToolsDialog component.
- Sanitize server-reported names into safe direct-tool prefixes (spaces/punctuation
become underscores) and use the same sanitization in the proxy resolver.
* fix(mcp): give the edit-dialog tool list a definite height
The Shadcn ScrollArea viewport is h-full, which requires a definite height
on the root. Using max-h-64 left the viewport at auto height, so the root
clipped the list at 256px without engaging its own scrollbar — tools past
the cutoff were unreachable. Use a fixed h-64 container (same pattern as
move-copy-dialog) so the inner scroll area scrolls independently of the
dialog's outer scroll.
* refactor(mcp): fixed header/footer with scrollable content in server config dialog
Mirror the skill config dialog layout: the edit dialog now keeps the header
and footer fixed and scrolls only the content area (ScrollArea flex-1
min-h-0 inside an h-[85vh] flex column). The tools list no longer has its
own nested ScrollArea — it flows inside the content-level scroll, so all
tools are reachable without a nested-scroll conflict. The create dialog
stays compact (URL-only) like the skill add dialog.
Verified headlessly that max-height-based ScrollArea roots do not engage
scrolling (viewport height resolves to auto), which is why the previous
nested list clipped; the flex-based definite-height pattern scrolls.
* feat(mcp): make the endpoint URL immutable after creation
Editing a server's URL effectively points it at a different server, and the
create flow is URL-only anyway — so the edit dialog no longer allows changing
it (delete + re-add covers endpoint changes). The URL is displayed read-only
next to the auto-detected name/description, and `url` was removed from the
update DTO and service; auth/transport changes still invalidate credentials
and cached tools.
* refactor(mcp): match MCP server cards to the skill card layout
The settings list cards now share the skill card structure: row 1 is the
server name (with direct-mode/transport badges) plus the three-dot menu,
row 2 is the description (falling back to 'No description provided') with
the endpoint URL and last error as small mono lines, and row 3 holds the
connection-status badge and updated date on the left with the permission
selector and the connect button at the right end.
* fix(mcp): stop card menu clicks from opening the edit dialog
The three-dot menu is portaled, so its item clicks bubble through React's
synthetic events to the card's onClick, which opened the edit dialog before
the delete confirmation appeared. Add e.stopPropagation() to every menu
item (matching the skill card pattern).
* fix(mcp): stop Refresh Tools from submitting/closing the config dialog
The Refresh Tools button lived inside the <form> and, like all buttons
without an explicit type, defaulted to type="submit" — clicking it ran the
form submit (save + close) instead of just refreshing. Set type="button".
Also per UX: while refreshing, the button is disabled with a spinning icon,
the per-tool enable/disable switches are disabled, and the success toast is
removed — the button reverting to its idle state is the success signal
(keep the error toast for failures).
* test(mcp): provide explicit redirectUri in McpOauthProvider test helper
* fix(mcp): scope tool registry to agent-assigned servers
The MCP tool registry was a process-global singleton keyed by server id,
so a chat agent could search, describe, and even call tools belonging to
MCP servers that were not assigned to it (e.g. disabled or unassigned
servers), as long as their metadata had been discovered at some point in
the process. This mirrors the disabled-skill bug: the read path never
checked per-agent enablement.
Replace the singleton with a fresh McpToolRegistry built per
buildAgentTools call and warmed from the persisted McpServer.tools DB
cache. Because the registry only ever contains the agent's assigned
servers, unassigned/disabled servers can neither be suggested, searched,
described, nor executed. Live discovery (lazy connect / connect action)
writes into the per-agent registry via ensureConnected(serverId,
registry?).
Adds test-first coverage: suggestions, search, describe and tool-call
leaks across an unassigned same-team server, staleness after
unassignment, and direct-tools scoping.
* feat(mcp): capture and surface server instructions + server-level refresh (#293)
* feat(mcp): capture and surface server instructions
- Add McpServer.instructions column, populated from getInstructions() at
connect time (legacy initialize + modern discover results)
- Surface instructions at three levels like pi-mcp-adapter: truncated
snippet in the mcp proxy tool description, preview in server listings,
and full text via mcp({ instructions: "name" })
- Fix misleading mcp({ server, connect }) usage line to mcp({ connect })
- Share one renderServerListing helper across list and connect modes
- Expose instructions in McpServerInfo DTO and WebUI config card
- Test server supports instructions; service + API tests updated
* refactor(mcp): server-level refresh endpoint, drop agent connect mode
- Replace POST /mcp/servers/:id/tools/refresh with POST /mcp/servers/:id/refresh
returning the refreshed McpServerInfo (server-level, not tools-level)
- Add McpService.refreshServer(): force-close the connection, re-discover and
persist tools, instructions, name/description and status; used by the refresh
endpoint, Test Connection, and post-OAuth completion (replacing discoverTools)
- Remove discoverTools entirely - refreshServer is the single discovery entry
point (testServer and completeAuth route through it)
- Remove refreshTools from UpdateMcpServerRequest/PATCH (dead param)
- Remove mcp({ connect: "name" }) from the agent proxy tool: lazy tool calls
still auto-connect, so agents are unaffected
- WebUI: remove the dialog 'Refresh Tools' button and create-mode background
auto-refresh; add a server-level Refresh button at the left end of the edit
dialog footer; i18n mcp_refresh_server (+ success message)
* feat(mcp): add in-flight request tracking and idle connection cleanup (#294)
* feat(mcp): add in-flight request tracking and idle connection cleanup
* feat(mcp): add background idle cleanup timer to McpService
* feat(mcp): update default background idle cleanup check interval to 30s
* feat(webui): improve MCP server card footer, instructions, and refresh dialog freeze (#295)
* feat(webui): improve MCP server list card, instructions dialog, and refresh freeze UI
* feat(webui): refine MCP card action menu items
* docs(webui): update direct tools mode description in i18n messages
* style(webui): refine Direct Tools Mode layout in MCP server config dialog
* feat(webui): remove tool search box from MCP server config dialog
* fix(mcp): make in-flight idle tracking test deterministic
The 'tracks in-flight requests and safely prunes idle connections' test
relied on a wall-clock sleep (10ms) to guess when the tool handler had
started executing on the server. On slow/loaded CI runners the HTTP round
trip through the real MCP client/server stack can exceed that window, so
resolveSlowTool() fired the no-op placeholder, the gated handler never
completed, and callPromise hung until the 5000ms test timeout.
The handler now signals via a deferred promise when it actually starts,
and the test awaits that signal (racing it against the call promise to
surface early failures) before asserting in-flight state and releasing
the gate. Also gives the test an explicit 10s timeout.
Disabling a skill in the agent config dialog only removed it from the
agent_skills join table, but the chat flow still advertised ALL team
skills to the model in the <available_skills> system prompt section and
the read_skill tool never checked per-agent enablement.
- getAgentChatContextActivity / getAgentAutofillContextActivity now only
return skills explicitly enabled for the agent (agent_skills), and
expose enabledSkillIds in the execution context.
- createReadSkillTool accepts enabledSkillIds and rejects read_skill
calls for skills not enabled for the agent, even if the model knows
the skill id from earlier conversation history.
- Added unit tests for context filtering (chat + autofill), tool-level
enforcement, and the zero-enabled-skills case.
* fix(webui): enable rename and delete actions in file detail top bar breadcrumb
* fix(webui): select only base file name when renaming from file detail breadcrumb
startDownload() used <a target="_blank" download> anchors, which made
Chrome open a new tab for every file that then auto-closed as soon as the
download started (Content-Disposition: attachment). Replace with hidden
iframes: downloads start in the background without opening tabs, and a
failed/expired link renders inside the hidden iframe instead of
navigating the SPA tab away.
* fix(core): use renamed asset name in download Content-Disposition
Downloads of renamed assets (e.g. test.png renamed to foo.png) previously
saved under the storage-key basename because the presigned URLs / local
file handler only set Content-Disposition: attachment without a filename,
and the HTML download attribute is ignored for cross-origin URLs.
- Add buildContentDisposition() helper that emits a sanitized
'attachment; filename=...; filename*=UTF-8...' header (RFC 5987) so
non-ASCII names survive.
- Thread an optional filename through S3Service.presign():
* S3 backend passes it via contentDisposition (response-content-disposition)
* Local backend carries it as a ?filename= query param consumed by the
/files serveStatic onFound handler
- getDownloadLinks / getDownloadUrl now pass the renamed asset name from
the DB, so all download flows (file browser, viewers, breadcrumb, share)
save files under the current asset name.
* fix(core): keep Content-Disposition header ASCII for non-ASCII names
Bun's Headers API rejects header values containing non-ASCII bytes, so a
download of a renamed file with a non-ASCII name (e.g. 幻境边界.pptx) threw
'TypeError: Header Content-Disposition has invalid value' on the local
storage backend.
Per RFC 5987, non-ASCII names belong only in the percent-encoded
filename* parameter. buildContentDisposition now emits the quoted
filename="..." fallback only when the name is pure ASCII; otherwise the
header carries just 'attachment; filename*=UTF-8''...' (all-ASCII, accepted
by Bun/Node Headers and all modern browsers).
* feat(transcode): add Gotenberg support for Office, HTML, CSV, and Markdown PDF proxies
* fix(transcode): use Gotenberg chromium markdown endpoint with index.html wrapper for markdown conversion
* test(transcode): add complete unit test coverage for HTML, CSV landscape/portrait, TXT, and Gotenberg fallback branches
* fix(gotenberg): force .md extension for Gotenberg markdown conversion endpoint compatibility
* refactor(gotenberg): standardize basic auth environment variables on GOTENBERG_BASIC_AUTH_USERNAME and GOTENBERG_BASIC_AUTH_PASSWORD
* test(e2e): add real Gotenberg testcontainers workflow E2E test for docx, pptx, and xlsx
* docs: add Gotenberg document PDF conversion documentation and update features in README
* docs(gotenberg): add Docker quickstart commands and remove obsolete GOTENBERG_USERNAME note
Split the monolithic test:e2e script into:
- test:e2e:app: fullstack web app e2e (playwright --project=app-chromium)
- test:e2e:webui: harness UI integration tests
(E2E_SKIP_APP_SERVER=true --project=harness-chromium)
Remove the now-unused test:e2e:ui and test:e2e:report scripts, and update
the AGENTS.md submission rules and e2e README to reference the new
commands.
Two issues made the watermarked share-link spec flaky on Firefox (and
silently mis-typed on every browser):
1. test.use inside a for loop leaks the last iteration's value to all
tests: both the 'image' and 'video' cases were actually running the
video flow. Scope each test.use inside a test.describe block.
2. Selecting the file by clicking the card body raced with the preview
area re-rendering (video sprite scrubber hover) and the Radix menu
exit animation, which could swallow the click on Firefox. Select via
the card checkbox instead, and wait for the menu to fully unmount
after Escape before the next interaction.
Verified stable: 6/6 on Firefox, 2/2 on Chromium, 2/2 on WebKit.
Document the two new share link security settings:
- Allow Download: hides all download affordances on the public pages and
rejects the download endpoint with 403 while keeping previews available.
- Watermark: per-share-link lazy watermarking — transcodes are only
triggered when the watermark is enabled on a share link (and again for
files added later), so Shumai never stores watermark copies for files
that are not shared through a watermarked link, saving storage.
Add web app e2e coverage for the frequently-combined watermark +
download-disabled share link configuration. For both watermarkable media
types (image and video) the test asserts:
- no download affordance on the public share list page (card menu,
selection bar) and the public file detail page (viewer control bar,
top-nav file dropdown; video also has no resolution submenu)
- the proxy loaded in the detail page is the watermarked version
(-watermark- S3 keys) and no original transcode key is requested
- the download-url endpoint rejects guest download attempts with 403
Extends the e2e share helpers with apiUpdateShare, apiEnableShareWatermark
and apiPublicDownloadUrl, and documents the new cases in the e2e README.
* feat(webui): remove original-file display fallbacks, use proxies only
WebUI now never uses the raw original file for display. All viewers
(image, video, compare, pdf) render exclusively from transcoded proxies;
when no proxy exists the viewer shows an unavailable placeholder instead
of falling back to media.original.
- getBestTranscode: drop the isRaw fallback candidates
- video-viewer / compare-video-pane: remove the synthetic 'Original'
resolution entry and the original.downloadUrl hasMedia gate
- image-viewer / compare-image-pane: remove original.downloadUrl default
- pdf-viewer: require pdfTranscode.url (no original fallback)
- download menus (video-control-bar, breadcrumb-nav): use
media.original.key for the 'Original' download option instead of the
isRaw transcode marker entry
Backend:
- transcode-video workflow no longer pushes the isRaw:true marker entry
into videoTranscodes (it pointed at the original file key)
- remove media.original.downloadUrl from the DTO, asset service mapping,
MediaInfo DB type, and transcode workflows/activities (it was only
consumed by the removed display fallbacks; downloads already use key
via the download-url endpoints)
Tests/fixtures updated; docs updated to state the proxy-only display
policy.
* refactor: remove isRaw field via data migration, clean legacy media JSON
Add a Prisma data migration that cleans legacy media JSONB on both
assets and watermark_files:
- drop the raw-original marker entries (isRaw: true) from
videoTranscodes / imageTranscodes — they pointed at the original file
key and were only used by the removed display fallbacks / download
menus, which now use media.original.key directly
- strip the isRaw field from remaining transcode entries
- drop original.downloadUrl (removed from the AssetInfo DTO)
The migration is generated via prisma migrate dev --create-only and
applied automatically by migrate deploy in production. It is guarded by
a test (packages/db/src/clean-media-migration.test.ts) that seeds legacy
media JSON, runs the migration SQL, and asserts the transform on both
tables.
With the data cleaned, remove the isRaw field entirely from the codebase:
- DTO schemas + ImageTranscode/VideoTranscode interfaces
- PrismaJson MediaInfo transcode types
- watermark activity isRaw skip (and its test), watermark proxy mapping,
agent-embedding transcode selection, webui transcode filters, and all
test fixtures
AGENTS.md: clarify that the no-manual-migration rule applies to DDL;
data migrations (pure SQL transforms, no schema change) are allowed
inside a --create-only generated migration and must be tested.
* feat(sharelink): add sharelink watermark support
* fix(sharelink): harden watermark feature against races, crashes, and tenant leaks
Addresses review findings on the sharelink watermark feature:
- fix(watermark): read back sharelink via transaction client so the
updateShareLinkWatermark response reflects the just-committed state
(global-client reads on a pooled connection miss uncommitted changes)
- fix(watermark): use prisma upsert on the config hash instead of
find-then-create to avoid P2002 races on concurrent enables
- fix(watermark): resolve nested symlinks when walking the share tree and
batch the walk level-by-level instead of one query per node
- fix(watermark): keep watermark tasks in flight from being purged by
orphan-config GC (avoids FK violations in completing activities)
- fix(transcode): re-transcode stale 'processing' watermark rows instead
of silently skipping (crash recovery); downscale embedded block images;
use pino structured logging; drop unused any casts
- fix(api): scope watermark-template list/create to the caller's teams,
reject teamless template creation, and make teamless templates
read-only; validate watermark config required when enabling (400);
map sharelink-not-found to 404
- fix(api): resolve symlink target ids once in public share children
endpoint instead of twice per item
- refactor(db): dedicated WatermarkFileStatus enum (migration included)
- test: activity unit tests (init crash recovery, image + video transcode
paths), concurrent upsert, nested-symlink resolution, API scoping and
validation, e2e video watermark path
* refactor(transcode): move sharp usage out of watermark activity into transcodeService
The transcode package does not declare sharp as a dependency (only
@shumai/core does), so importing it directly in the watermark activity
broke CI typecheck for the transcode package.
- add transcodeService.renderSvgToPng, downscaleImageToPng and
compositeOverlayToWebpFile encapsulating the sharp logic
- watermark activity now calls these methods instead of importing sharp
- activity tests mock the new methods (no sharp/real fixture needed)
- add unit tests for the new transcodeService helpers
* feat(watermark): restrict watermark transcode to existing image and video transcodes
* refactor(transcode): reuse transcodeVideo in watermark activity
* refactor(watermark): use Bun.hash instead of node crypto for watermark config hash
* refactor(watermark): move computeWatermarkConfigHash from dtos to core watermark-hash module
* refactor(api): delegate watermark DB queries from public-share route to watermarkService
* refactor(share): replace dynamic watermarkService import with static import
* feat(watermark): add waitForWatermarkFileActivity for in-flight watermark transcoding tasks
* refactor(asset): replace dynamic watermarkService import with static import
* test(watermark): harden purgeOrphanWatermarkConfigs coverage
Replace the single happy-path GC test with a dedicated suite covering:
- orphan config with proxies: purged, watermarkFiles removed, all four
proxy keys deleted from S3 while the original key is preserved
- config still referenced by a share link is not purged
- config with pending/processing transcode_watermark tasks is not purged
(in-flight guard), but completed tasks do not block purging
- pending tasks of other types (transcode_video) do not protect a config
- an S3 delete failure is non-fatal: remaining keys still deleted and the
config is still purged
- mixed run: only the orphan config is purged while a referenced one
survives (exact count assertion)
* feat(webui): add sharelink watermark editor and sidebar controls
* style(webui): enlarge watermark editor dialog and allocate 2/3 width to preview column
* fix(webui): override sm:max-w-lg in DialogContent to allow full 1400px width
* fix(webui): stop block click propagation and remove preview mode toggles
* fix(webui): store watermark block size as fraction of canvas width
The editor stored block.size in pixels (slider 10-120px, preview
fontSize: `${block.size}px`) while the SVG generator
(generateWatermarkSvg) interprets size as a fraction (0..1) of canvas
width. A 36px block therefore rendered as a ~36x-canvas glyph covering
the entire image.
- Keep backend fraction semantics (canvasWidth * size) so watermarks
scale proportionally with output resolution across proxies
- Editor: defaults (text 5%, image 15%), slider ranges (1-12% / 2-50%)
and number input now work in percent, storing a fraction
- Preview: scales blocks by the measured canvas width so WYSIWYG
- DTO schema: size constrained to (0..1) to enforce the contract
- Update DTO/editor tests to fraction sizes
* feat(webui): add single-row aspect ratio selector and 5 predefined background colors to watermark editor preview
* feat(webui): update aspect ratio to multi-toggle group and enable dot-grid pattern for all background colors
* fix(webui): update background swatches to distinct colors and share PRESET_BACKGROUND_OPTIONS array
* fix(webui): hide redundant absolute close button on watermark editor dialog header
* fix(transcode): skip raw video transcode when generating watermark proxies
The raw marker entry (isRaw: true) in videoTranscodes points at the
original source file and is only used to offer the 'Original' playback
option. It is not a transcoded proxy, so watermarking it produced a
redundant full-resolution transcode (the most expensive pass) plus an
unneeded S3 upload.
Skip entries with isRaw: true in transcodeWatermarkMediaActivity, and
cover the behavior with a unit regression test plus an e2e assertion
that exactly one watermark proxy is produced for a video asset.
* feat(webui): replace watermark image asset ID with 1MB file uploader
* fix(webui): correct watermark editor preview block sizing
The preview canvas width was measured in a useEffect([open]) that ran
before the Radix Dialog had mounted its content, so canvasRef.current was
still null, canvasWidthPx stayed 0, and every block silently fell back to
the hardcoded 640px width (block.size * (canvasWidthPx || 640)). A 100%
size therefore rendered at ~640px on an ~876px canvas (~72%), making the
preview smaller than the real watermarked output for both image and text
blocks.
- Measure the canvas via a ref callback (fires exactly when the node
mounts, regardless of dialog mount timing) and keep a ResizeObserver
attached for aspect-ratio/window changes.
- Use clientWidth (padding-box width) instead of getBoundingClientRect(),
which includes the dialog's zoom-in-95 animation transform.
- Drop max-w-full/max-h-full on the preview <img> so logos scale up to
the block size like the backend SVG rendering does, instead of being
capped at their natural pixel size.
* fix(webui): keep watermark editor preview aspect ratio and restore config scroll
- Compute the preview canvas size with a JS contain-fit (ResizeObserver on the
frame) instead of relying on CSS aspect-ratio + w-full/max-h-full, which
broke the selected ratio on wide-and-short (or tall-and-narrow) windows.
- Add min-h-0 to the config ScrollArea so it shrinks to the column's remaining
height instead of growing to its content height, restoring the scroll range.
* fix(webui): remove toast notification on template load
* fix(share): serve per-file watermark proxies in public share APIs
Stop gating watermark replacement on the share-link-wide watermarkStatus.
Public share endpoints now resolve each asset's own completed watermark
proxy and:
- replace videoTranscodes/imageTranscodes with the watermarked entries
when the watermark file is completed;
- empty those arrays when no completed proxy exists yet (still
transcoding or failed) so original, unwatermarked media is never
exposed while a newly added file is being processed;
- leave non-video/image media (pdf, audio, ...) untouched, since they
never receive watermark transcodes.
This removes the previous behavior where adding a file to a ready
watermark share flipped the whole share link to processing and served
original media for every file until the new transcode finished.
Adds API tests for detail/children endpoints (ready, in-flight, pdf, and
watermark-disabled cases) and a workflow E2E test for adding an asset to
an already-ready watermark sharelink.