loop-repo

name	loop-repo
description	Maintain a repository on a recurring heartbeat by inspecting issues, pull requests, CI, dependencies, and repository-local priorities, then directing bounded work to fresh Codex tasks. Use when the user asks Codex to maintain a repo, run recurring engineering triage, or wake periodically and dispatch useful work.

Loop Repo

Run a conservative maintainer heartbeat. Find useful machine-checkable work, delegate it, and leave the repository cleaner than the previous wakeup.

Workflow

1. Read repository instructions and identify its active branch, local notes, roadmap, open issues and PRs, recent CI, and existing maintainer state.
2. Define the cadence, priority sources, verification gates, concurrency limit, merge authority, and terminal condition. Default to one active implementation lane and one review lane.
3. On each wakeup, refresh live state before choosing work. Prefer already-started or blocking work over generating new work.
4. Select only bounded, in-scope tasks with objective verification. Avoid vague product work, architecture rewrites, auth, payments, deployments, and destructive changes unless explicitly authorized.
5. Create fresh user-visible Codex tasks in isolated worktrees. Keep implementation and independent review separate.
6. Track task ids, branches, PRs, attempts, last observation, and next action. Do not dispatch duplicate work for unchanged state.
7. Carry owned PRs through checks and review, using $loop-pr behavior. Respect the configured merge gate.
8. End or archive the heartbeat when the terminal condition holds. Escalate repeated blockers with evidence.

Return a short report after meaningful action. Archive no-op wakes when the product supports it.