This repository was archived by the owner on Feb 6, 2026. It is now read-only.
Repository navigation
This repository was archived by the owner on Feb 6, 2026. It is now read-only.
Failed executions can appear as successful #7
Description
Activity
Hi, this has happened to me as well.
Although this review was just a few lines of code, so it does not seem to be a problem related to how big the PR is.
Starting review for PR #71 Repository: ####### Action: opened Environment variable GITHUB_WEBHOOK_SECRET not found, keeping placeholder Fetching diff content from GitHub API... Retrieved diff content (2359 chars) Calling reviewDiff() for job action-1764001324044-71 [Amp] Starting thread... [Amp] Started thread: T-dbe51b33-7ef7-4862-a3e3-dbb4dae5741a [Amp] Review completed successfully Error starting thread: Error: Amp review failed [Amp] Iterator completed Review job action-1764001324044-71 failed with exception: Error: Failed to start thread: Error: Amp review failed at reviewDiff (file:///app/dist/review/reviewer.js:70:15) at process.processTicksAndRejections (node:internal/process/task_queues:95:5) at async processReview (file:///app/dist/github/process-review.js:73:30) at async main (file:///app/dist/bin/review-action.js:26:9) Review completed successfully
Ok so this was blocking multiple Amp reviews in one of my repos due to a specific edge case I identified:
When the AI agent uses only toolbox functions (
leave_inline_comment,leave_general_comment) without returning additional text,resultbecomes an empty string. Since empty strings are falsy in JavaScriptt, the check failed even though the review succeeded (thread was created, comments were written to the JSONL file).I checked the internals, downloaded the Docker image locally, reproduced the bug, and confirmed the fix resolves the issue.
Created PR #8 with the fix.
Thanks!
We notice that things can fail if the PR is a big one, but the problem is that the workflow is not marked as failed in these cases.
In fact, it shows up as successful:
Is this intentional?