NemoClaw: Skill for Codex

Continuously Fix E2E Failures on Main

Continuously inspect automatic E2E results for main. Coordinate ownership and merge decisions through GitHub.

Set the Operating Rules

1. Start without a scheduled end time. Do not infer an end time from local time, a shift change, a passing run, or an empty queue.
2. Keep release operations out of scope. Never change, retag, publish, or otherwise touch a release, tag, or release artifact during this workflow. Route release work to the existing release workflow.
3. Confirm maintainer authorization. Merge only when the request grants it. Otherwise, leave the PR approval-ready and continue the loop.
4. Check Git and GitHub access. Follow Git and GitHub Access Hard Stop on access failure.

Do not declare success or end the loop because the queue is empty or the newest run passes. Wait for the next automatic main result.

Keep the Queue

Read Queue and Ownership before the first scan. Keep one table grouped by root cause:

Root cause	Run and jobs	State	Owner and PR	Next action

Use only these states: unclaimed, active, waiting-ci, waiting-review, approval-ready, merged, obsolete, and blocked.

Track each observed workflow run by run ID, attempt, status, conclusion, and job set. Re-read an in-progress or queued run when its state changes. Do not reanalyze an unchanged completed run.

Run the Loop

Repeat these steps continuously while the loop remains authorized:

1. Refresh origin/main. List automatic E2E runs for that commit SHA and later main commit SHAs.
2. Inspect only new or changed runs. Load nemoclaw-maintainer-classify-ci-failure to read one failed job's bounded logs and optional retained artifact before identifying the earliest actionable product, test, workflow, runner, or cleanup failure.
3. Group failures that share the same causal signature. Do not equate a job name with a root cause.
4. Reconcile each group with open PRs before editing. If another maintainer owns it, record that PR and take the next unowned group.
5. Prefer a peer E2E maintenance PR that needs review or a final merge decision before starting another fix.
6. Select one unowned root cause. Claim it before the product fix with a draft PR whose initial diff contains evidence for only that root cause.
7. Work on only that root cause. Add the diagnostic or regression evidence that should have caught an escaped defect.
8. When the PR waits for CI or peer review, stop editing it. Review a peer PR or take the next unowned root cause. Edit only one fix at a time.
9. Revisit waiting and blocked groups during each scan. Rescan after a run, check, review, PR, or merge state changes. Rescan for each new automatic main result.