Back to feed
Dev.to
Dev.to
7/1/2026
Our AI autopilot merged nothing for 3 days — the culprit was a review 'freshness' check

Our AI autopilot merged nothing for 3 days — the culprit was a review 'freshness' check

Short summary

Codens' AI autopilot system froze for three days when a defensive freshness check in review polling logic incorrectly filtered out standing approvals after system recovery reset the reference timestamp. The fix: stop asking 'is this review new?' and instead adopt the external system's own semantics by querying the current effective review state, matching GitHub's own behavior.

  • •AI autopilot halted PR merges for three days due to a freshness check filtering approved reviews after system recovery
  • •System restart reset the reference timestamp used to filter reviews, orphaning all standing approvals permanently
  • •Fix: query the effective review state per reviewer, mirroring GitHub's own semantics rather than tracking review newness

Generated with AI, which can make mistakes.

#ai-agents#ai-tools
Read full article at Dev.to

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

Nobody wants to review the robot's 600-line pull request
Dev.to
The original title is: "Using a Second AI to Review Every Commit: Triage Lessons from CLAUDE.md"
Dev.to
State-bound intent: solving the stale state problem in AI agent workflows
Dev.to
The original headline is: "We Let AI Write a Third of Our Code. Here's the Review Process That Kept Us Sane."
Dev.to