Fix agent CLI stuck with no timeout on stalled LLM streams #147
Conversation
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
Adding CLAUDE.md with task information for AI processing. This file will be removed when the task is complete. Issue: #146
Contributor
Author
🚨 Solution Draft FailedThe automated solution draft encountered an error: 📎 Failure log uploaded as Gist (790KB)
|
Contributor
Author
|
🤖 AI Work Session Started Starting automated work session at 2026-01-30T23:08:57.801Z The PR has been converted to draft mode while work is in progress. This comment marks the beginning of an AI work session. Please wait working session to finish, and provide your feedback. |
Deep analysis of the 2h10m hang incident including: - Complete timeline reconstruction from logs - Root cause: missing streamText timeout configuration - Contributing factors: no session inactivity watchdog, unref'd intervals - Proposed solutions using AI SDK's built-in timeout options Refs #146 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Root cause: The streamText() call had no timeout configuration, so when the upstream LLM API connection stalled (TCP alive but no data), the process waited indefinitely with no recovery mechanism. Changes: - Add chunkMs (2min) and stepMs (10min) timeouts to streamText() call using the AI SDK's built-in timeout option - Add configurable env vars: AGENT_STREAM_CHUNK_TIMEOUT_MS (default 120s) and AGENT_STREAM_STEP_TIMEOUT_MS (default 600s) - Add .unref() to setInterval in continuous mode to allow natural exit - Add tests for stream timeout configuration Fixes #146 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
konard
changed the title
This reverts commit 60770ff.
Contributor
Author
🤖 Solution Draft LogThis log file contains the complete execution trace of the AI solution draft process. 💰 Cost estimation:
Now working session is ended, feel free to review and add any feedback on the solution draft. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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
Fixes #146 - Agent CLI hangs indefinitely when the LLM API stream stalls (TCP connection alive but no data flowing).
Root Cause
The
streamText()call injs/src/session/prompt.tshad no timeout configuration. When the upstream LLM API connection stalled (TCP alive, no data), the process waited indefinitely — in the reported incident, for 2 hours and 10 minutes until manual CTRL+C.The Vercel AI SDK (v6) provides a built-in
timeoutoption withchunkMs(timeout between stream chunks) andstepMs(timeout per LLM step), but these were not being used.Changes
js/src/session/prompt.ts: Addtimeout: { chunkMs, stepMs }to thestreamText()callchunkMs: 2 minutes (detects stalled streams — no new chunk within this duration aborts the call)stepMs: 10 minutes (max time for each individual LLM step)js/src/flag/flag.ts: Add configurable environment variables:AGENT_STREAM_CHUNK_TIMEOUT_MS/LINK_ASSISTANT_AGENT_STREAM_CHUNK_TIMEOUT_MS(default: 120000)AGENT_STREAM_STEP_TIMEOUT_MS/LINK_ASSISTANT_AGENT_STREAM_STEP_TIMEOUT_MS(default: 600000)js/src/cli/continuous-mode.js: Add.unref()tosetIntervalpolling loops to allow natural process exitjs/tests/stream-timeout.test.js: 6 tests for timeout configuration (defaults, env var overrides, prefix priority)docs/case-studies/issue-146/: Complete case study with timeline reconstruction, root cause analysis, and proposed solutionsHow It Prevents the Issue
When a stream stalls, the
chunkMstimeout will trigger after 2 minutes of inactivity. This causes the AI SDK to abort the stream, which surfaces as an error in the processor. The existing retry logic (forTimeoutErroradded in PR #143) will then automatically retry the request.Test Results
All 44 timeout-related tests pass (6 new + 22 existing MCP timeout + 16 timeout retry tests).
Fixes #146