[pull] main from czlonkowski:main - #66
Merged
Merged
Conversation
…rkflow (#777) (#786) * fix(workflow): strip duplicate activeVersion, add mode='active' (#777) n8n's draft/publish model returns a nested activeVersion alongside the draft on every workflow GET, duplicating the live graph's nodes and connections. On active workflows this pushed responses past Claude Code's default 25k-token MCP cap, so the host persisted results to a temp file the model could not read. handleGetWorkflow (mode 'full') and handleGetWorkflowDetails (mode 'details') now strip the heavy activeVersion payload while preserving the activeVersionId pointer. A new mode 'active' exposes the published graph as a single-shaped response (nodes/connections come from the live version, with publishedAt and versionName at the top level), returning NO_ACTIVE_VERSION for workflows that were never activated. As defense in depth for genuinely huge workflows, the n8n_get_workflow tool definition now declares _meta["anthropic/maxResultSizeChars"]: 500000 per the Claude Code MCP spec. UIAppRegistry.injectToolMeta now spread-merges instead of clobbering so per-tool _meta keys survive UI metadata injection. Reported by @nepalez. Conceived by Romuald Członkowski - https://www.aiadvisors.pl/en Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * review: simplify helper, prune type, rename publishedAt, tighten size cap Apply code-simplifier, code-reviewer, and Copilot feedback on PR #786: - stripActiveVersion: drop unnecessary generic and dead falsy guard. Plain (workflow: Workflow) => Workflow is fine — both call sites pass a full Workflow. - ActiveWorkflowVersion: prune speculative fields (versionId, workflowId, authors, autosaved, description, updatedAt, workflowPublishHistory) that no consumer reads. Keep an open index signature so the additional keys on the wire don't cause type errors. - handleGetWorkflowActive: rename publishedAt to versionCreatedAt — the value is activeVersion.createdAt, which is the version row's creation timestamp, not the publish-event time. n8n's workflowPublishHistory carries the actual activation event ~1s later. Don't claim equivalence. Also fall back to null when the source is missing. - maxResultSizeChars: 500_000 → 450_000 to leave headroom for the MCP/JSON-RPC envelope wrapping our payload. - Add regression tests: mode='full' must return the DRAFT nodes/connections (not the published ones) after the strip; pre-draft/publish n8n versions (no activeVersion key at all) must pass through unmangled. No behavior change to the size-fix and mode='active' contracts. Conceived by Romuald Członkowski - https://www.aiadvisors.pl/en Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * test: cover handleGetWorkflowActive error paths and mode dispatch Codecov flagged 25 lines of patch coverage gap (70.93% → target). The gap was concentrated in two places: - handleGetWorkflowActive's catch ladder (Zod / N8nApiError / generic) — every other handler in this file has matching error tests but the new one didn't. Add four: invalid input, N8nApiError mapping, generic Error fall-through, and a defaults-to-null check on versionCreatedAt / versionName when the active version row carries no name or timestamp. - src/mcp/server.ts's 'active' case in the n8n_get_workflow switch is trivial wiring but had no unit-level coverage. Add a focused dispatch test file that mocks handlers-n8n-manager via vi.hoisted and asserts mode='active' routes to handleGetWorkflowActive (and mode='details' and default route to their respective handlers as regression guards). +7 tests, all green. No production code change. Conceived by Romuald Członkowski - https://www.aiadvisors.pl/en Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix: address Copilot follow-up review - CHANGELOG: correct the documented size cap (450000, not 500000) so the release notes match the implementation after the earlier reviewer pass. - handleGetWorkflowActive: fall back to workflow.nodes/connections when workflow.active is true but no activeVersionId/activeVersion is present. This makes mode='active' usable across n8n versions — pre-draft/publish n8n returns no activeVersion field at all, and on those instances the workflow body itself IS the running graph. The same branch covers the rare orphan case in newer n8n where activeVersionId got nulled but the workflow is still flagged active. NO_ACTIVE_VERSION now means "inactive and never published" specifically. - Tests updated: the "active:true, no activeVersionId" path now asserts the fallback returns workflow.nodes; a separate "active:false, no activeVersionId" test pins the genuine never-activated case to NO_ACTIVE_VERSION. Added an orphaned-on-inactive test for completeness. - Tool doc updated to describe the fallback and the narrower NO_ACTIVE_VERSION semantic. - Force-track dist/mcp/ui/registry.* so Copilot stops flagging the gap — the file is regenerated by the release pipeline (npm run build:all before npm publish), but committing it removes the false-positive. Conceived by Romuald Członkowski - https://www.aiadvisors.pl/en Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
…rebased #783 + #784) (#787) * fix(http): preserve shared tenant sessions (cherry picked from commit 7e9e274) * fix(http): return 404 for stale sessions (cherry picked from commit e855570) * chore(dist): regenerate sourcemaps after cherry-picks * test(http-session): tighten shared-mode assertion The cherry-picked shared-mode regression test used `not.toHaveBeenCalledWith(400)`, which would also pass on 500/etc. Mirror the instance-mode test's tighter pattern: capture the existing transport reference, then assert it remains the same object and its `close` mock was never invoked. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(changelog): document HTTP session fixes for 2.53.0 Add entries for #783 (preserve shared tenant sessions) and #784 (404 for stale sessions) under the in-flight 2.53.0 section, both crediting @LevSky22. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs(changelog): drop dead 'malformed format' phrasing isValidSessionId() accepts any non-empty string, so the "malformed format" 400 branch is unreachable in practice. The changelog text now mirrors the implemented behavior: 400 for missing session IDs on non-initialize requests, 404 for unknown non-empty session IDs. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: levsky22 <56281588+LevSky22@users.noreply.github.com> Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )