auto-run
Autonomous dispatch-reconcile loop for batch task processing. Use with /auto-run --through <id> to execute tasks unattended. Requires beads tasks to exist. Supports --resume for checkpoint recovery and --skip-milestone-review.
npx skills add majiayu000/claude-skill-registry --skill auto-run --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
# Autonomous Orchestrator Loop: $ARGUMENTS You are an autonomous orchestrator. You dispatch workers, handle their completions, reconcile results, and dispatch newly unblocked tasks — repeating until all work is done or limits are reached. ## Section 1: Argument Parsing Arguments: `$ARGUMENTS` Parse the following flags: - `--max-batches N` — Stop after N dispatch rounds (default: unlimited) - `--max-hours H` — Stop after H hours (default: unlimited) - `--max-concurrent N` — Max parallel workers per batch (default: 3) - `--dry-run` — Show what would be dispatched without acting - `--resume` — Resume from checkpoint (skip orient) - `--skip-milestone-review` — Skip the milestone review phase after tasks complete - `--milestone-review-iterations N` — Max milestone review iterations (default: 5) - `--through <task-id>` — Complete everything needed to finish this task, then stop - `--epic <epic-id>` — Complete all tasks within this epic, then stop - `--only <id1> <id2> ...` — Only dispatch these specific tasks (and their blockers) ## Section 2: Initial Setup ### Check for Checkpoint Read checkpoint at `docs/auto-run-checkpoint.json` (resolve from main repo root via `git worktree list | he
- Section 1: Argument Parsing
- Section 2: Initial Setup
- Check for Checkpoint
- 2.1 Resolve Scope
- First Dispatch
- Section 3: Main Loop (Event-Driven)
- Step A — Identify Completion
- Step B — Reconcile
- Step C — Check Limits
- Step D — Dispatch Next Batch
- Step E — Context Self-Monitoring
- Section 4: Completion
- Step 1.5: Milestone Review Phase
- Checkpoint Schema
ls docs/session_summaries/<task-id>*.txt 2>/dev/null bd show <task-id> # get BLOCKED BY list Recursively resolve each blocker's blockers bd list --all | grep <epic-id> Check the session summary for frontend file extensions grep -E '\.(tsx|jsx|vue|svelte|html|css|scss)' docs/session_summaries/<task-id>*.txt 2>/dev/null bd create --title="Investigate: <task-id> failed" --type=task --priority=1 --parent <epic-id> git branch -r --list 'origin/milestone/*' | sort -V | tail -1
What does the auto-run skill do?
Autonomous dispatch-reconcile loop for batch task processing. Use with /auto-run --through <id> to execute tasks unattended. Requires beads tasks to exist. Supports --resume for checkpoint recovery and --skip-milestone-review.
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill auto-run --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.