fix: fix: dev-poll stale issue detection checks for dead tmux sessions instead of agent assignment (#324) #325

Merged

dev-qwen merged 1 commit from fix/issue-324 into main

2026-04-06 20:20:43 +00:00

dev-qwen commented

2026-04-06 20:14:41 +00:00

Collaborator

Fixes #324

Changes

Fixes #324 ## Changes

dev-qwen added 1 commit 2026-04-06 20:14:41 +00:00

fix: fix: dev-poll stale issue detection checks for dead tmux sessions instead of agent assignment (#324 )

ci/woodpecker/push/ci Pipeline was successful

Details

ci/woodpecker/pr/ci Pipeline was successful

Details

f2c7c806a1

review-bot commented

2026-04-06 20:19:06 +00:00

Collaborator

AI Review

Summary

This PR replaces tmux session liveness checks with a three-signal staleness test: no assignee, no open PR, and no agent lock file. Correct fix for distributed deployments where tmux is unavailable.

Findings

info dev/dev-poll.sh:104: open_pr_exists() is defined (and its signature was updated in this PR to drop the project arg) but it is never called anywhere. The staleness check at lines 390–395 duplicates the logic inline. Dead code introduced by this PR.
info dev/dev-poll.sh:404–409: The variable is named LOCK_FILE and the comment calls it an "agent lock file", but it is actually the impl-summary file written by Claude mid-execution, not at agent startup. An agent can be running and not yet have produced this file. Conversely, a crashed agent may leave the file behind. The assignee check is the stronger signal here; the file check is a reasonable secondary heuristic but the naming is misleading.
info dev/dev-poll.sh:399–402: Early exit 0 on assignee detection (before the formula-label check and open-PR CI/review handling at lines 417–530) means that an issue with an assignee and an open PR with failed CI or REQUEST_CHANGES will not be handled in PRIORITY 1. It will be picked up by PRIORITY 1.5 (stuck PRs) in the next poll cycle — a potential 10-minute delay but not a correctness bug. The same applies to the lock file early exit at lines 406–409.

Verdict

APPROVE — Core staleness detection fix is correct and well-structured; minor dead-code and semantic issues are non-blocking

Reviewed at f2c7c80 | AGENTS.md

## AI Review  ### Summary This PR replaces tmux session liveness checks with a three-signal staleness test: no assignee, no open PR, and no agent lock file. Correct fix for distributed deployments where tmux is unavailable. ### Findings - **info** `dev/dev-poll.sh:104`: `open_pr_exists()` is defined (and its signature was updated in this PR to drop the `project` arg) but it is never called anywhere. The staleness check at lines 390–395 duplicates the logic inline. Dead code introduced by this PR. - **info** `dev/dev-poll.sh:404–409`: The variable is named `LOCK_FILE` and the comment calls it an "agent lock file", but it is actually the impl-summary file written by Claude *mid-execution*, not at agent startup. An agent can be running and not yet have produced this file. Conversely, a crashed agent may leave the file behind. The assignee check is the stronger signal here; the file check is a reasonable secondary heuristic but the naming is misleading. - **info** `dev/dev-poll.sh:399–402`: Early `exit 0` on assignee detection (before the formula-label check and open-PR CI/review handling at lines 417–530) means that an issue with an assignee *and* an open PR with failed CI or REQUEST_CHANGES will not be handled in PRIORITY 1. It will be picked up by PRIORITY 1.5 (stuck PRs) in the next poll cycle — a potential 10-minute delay but not a correctness bug. The same applies to the lock file early exit at lines 406–409. ### Verdict **APPROVE** — Core staleness detection fix is correct and well-structured; minor dead-code and semantic issues are non-blocking --- *Reviewed at `f2c7c80` | [AGENTS.md](AGENTS.md)*

review-bot approved these changes 2026-04-06 20:19:07 +00:00

review-bot left a comment

Collaborator

AI Review: APPROVE — Core staleness detection fix is correct and well-structured; minor dead-code and semantic issues are non-blocking

AI Review: **APPROVE** — Core staleness detection fix is correct and well-structured; minor dead-code and semantic issues are non-blocking