fix(server): explicit manual-resume hint when web UI rejects/approves a non-web-parent run#1523
Open
blankse wants to merge 1 commit intocoleam00:devfrom
Open
Conversation
… a non-web-parent run Background When a workflow run was started from the CLI (or any non-web platform) and is then approved/rejected via the Web UI, `tryAutoResumeAfterGate` correctly skips dispatch — `dispatchToOrchestrator` is wired to the web adapter and would misroute Slack/Telegram/CLI parents. The skip is intentional. The bug is in what the user sees afterwards: the API response said only "Send a message to continue" or "On-reject prompt will run on resume", which is meaningless to a web-UI user whose run was started from a terminal. The Web UI dropped success messages entirely (only `setActionError` rendered), so even the vague hint never reached the user. The run sits in `failed` status with `metadata.rejection_reason` populated and no clear next step. Closes coleam00#1522. Change `tryAutoResumeAfterGate` now returns a structured discriminated union instead of a plain boolean: { resumed: true } | { resumed: false; reason: 'no_parent' | 'no_platform_conv' | 'non_web_parent' | 'dispatch_failed' } The four `reason` values mirror the existing log-event guard branches one-to-one. A new `manualResumeMessage()` helper constructs the user-facing hint per reason — CLI-only command for `no_parent`, both options for chat parents (`non_web_parent`), CLI-with-context for `no_platform_conv`, and "dispatch failed" for the catch-branch. Every non-resumed branch now names the exact `archon workflow resume <id>` command so the user has an actionable next step. The `/api/workflows/runs/:runId/resume` endpoint also surfaces the same explicit command instead of "Re-run the workflow to auto-resume". Tests Existing approve/reject auto-resume tests updated to assert the new explicit hint format. Added regression coverage for all four non-resumed branches on both approve and reject (eight new tests), each verifying the run-id-substituted CLI command appears in the response. The existing happy paths (web-parent dispatch, full-cancel, max-attempts) are untouched. Behavior change is response-text only — the dispatch decision logic, log events, and DB writes are byte-identical.
📝 WalkthroughWalkthroughAuto-resume failures during workflow reject/approve now return structured error reasons ( Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~22 minutes Possibly related issues
Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Review rate limit: 7/8 reviews remaining, refill in 7 minutes and 30 seconds.Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
🧹 Nitpick comments (1)
packages/server/src/routes/api.ts (1)
1151-1167: ⚡ Quick winAdd exhaustive
defaultguard to theswitchto prevent silentundefinedon future union extension.All four current members of
reasonare handled, but without adefault: neverbranch the function will returnundefinedat runtime if the union is extended — TypeScript only catches this reliably whennoImplicitReturns: trueis active in tsconfig.✅ Proposed fix
case 'dispatch_failed': return `Auto-resume dispatch failed. Run ${cliCommand} from a terminal to ${verb}.`; + default: { + // Exhaustive check — forces a compile error if the union grows without a matching case. + const _exhaustive: never = reason; + return `Run ${cliCommand} from a terminal to ${verb}.`; + } }🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@packages/server/src/routes/api.ts` around lines 1151 - 1167, The switch over the variable `reason` currently handles four cases but lacks an exhaustive `default`, which can lead to returning undefined if the `reason` union is extended; modify the switch (the block that references `reason`, `cliCommand`, and `verb`) to include a `default` branch that enforces exhaustiveness—e.g., assign `reason` to a `never`-typed variable or call a shared `assertUnreachable(reason)` helper and then throw a clear Error (or return a safe fallback message) so the compiler/runtime will catch any future unknown `reason` values.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@packages/server/src/routes/api.ts`:
- Around line 1151-1167: The switch over the variable `reason` currently handles
four cases but lacks an exhaustive `default`, which can lead to returning
undefined if the `reason` union is extended; modify the switch (the block that
references `reason`, `cliCommand`, and `verb`) to include a `default` branch
that enforces exhaustiveness—e.g., assign `reason` to a `never`-typed variable
or call a shared `assertUnreachable(reason)` helper and then throw a clear Error
(or return a safe fallback message) so the compiler/runtime will catch any
future unknown `reason` values.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: f57de0a2-6e16-4a15-b160-e59c374f7ec8
📒 Files selected for processing (2)
packages/server/src/routes/api.tspackages/server/src/routes/api.workflow-runs.test.ts
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 join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Summary
tryAutoResumeAfterGatecorrectly skips dispatch — but the API response only said "Send a message to continue" / "On-reject prompt will run on resume", and the Web UI rendered no success message at all. The run lands infailedstatus withmetadata.rejection_reasonpopulated and the user has no idea how to proceed.df-implement-with-preview-fastafter a real reject — the workflow died silently mid-iteration with a one-paragraph rejection comment that the on_reject prompt was supposed to consume. Recovery required reading the DB schema to discoverarchon workflow resume <id>exists.tryAutoResumeAfterGatenow returns a structured{ resumed } | { resumed: false; reason }discriminated union. Each non-resumed branch produces a tailored next-step hint that names the exactarchon workflow resume <run-id>command. The Resume endpoint message gets the same upgrade.parent_conversation_id-based cross-adapter guard, log events, DB writes — all byte-identical. No Web UI changes (toast/banner work is called out as a separate follow-up in bug(web): Reject + Resume buttons dead-end silently for non-web-parent runs #1522).UX Journey
Before
After
Architecture Diagram
Before
After
Connection inventory:
rejectWorkflowRunRoutetryAutoResumeAfterGateAutoResumeResultdiscriminatorapproveWorkflowRunRoutetryAutoResumeAfterGaterejectWorkflowRunRoutemanualResumeMessageresumed === falseapproveWorkflowRunRoutemanualResumeMessageresumed === falseresumeWorkflowRunRoutetryAutoResumeAfterGate→ log event namesLabel Snapshot
risk: lowsize: Sserverserver:api-routesChange Metadata
bugserverLinked Issue
Validation Evidence
api.workflow-runs.test.tsadds 8 new tests covering all four non-resumed branches × {approve, reject} and asserts each response message contains the substituted run id and the literalarchon workflow resume. Existing happy-path tests for web-parent dispatch unchanged. The pre-existingloader.test.ts/dag-executor.test.tsfailures ondevreproduce on a clean checkout in this WSL2 environment without my changes — unrelated.bun run lintwithout raised heap OOMs in WSL2 ondev(also unrelated). Ran withNODE_OPTIONS='--max-old-space-size=8192'and got clean.Security Impact
Compatibility / Migration
successboolean and HTTP status are unchanged. No callers parse the message content beyond display.Human Verification
archon workflow resume <id>, no orchestrator dispatch, run set tostatus='failed'withrejection_countincremented.archon workflow resume <id>, no orchestrator dispatch.archon workflow resume <id>.archon workflow resume <id>.dispatch_failedbranch and containsarchon workflow resume <id>.Side Effects / Blast Radius
packages/server/src/routes/api.ts(3 endpoints: approve / reject / resume).archon workflow resume,originating thread,no longer available,Auto-resume dispatch failed. The Web UI does not do that today.api.workflow_*_auto_resume_*) are unchanged → existing dashboards keep working.Rollback Plan
dev. No state migration, no UI dependency.Risks and Mitigations
archon workflow resume <id>string. That's the intended behavior here — but if the binary name ever changes, this string needs to follow. Mitigation: the literal lives in one helper (manualResumeMessage); a singleFind Usagescovers it.archon workflow resume <id>,originating thread,no longer available) rather than the full sentence.🤖 Generated with Claude Code
Summary by CodeRabbit
archon workflow resume <runId>) to all resume-related messages for manual recovery.