agentleFS
dannys-code-cornerdannys-code-corner/incan/.agents/skills/fleet-audit/SKILL.md

fleet-audit

Skill73 starsChanged 5 days agoApache-2.0View on GitHub

Report which local git worktrees under tmp/ are safe to close, need a human look, or are explicitly protected. Use when the user says /fleet-audit, asks what worktrees can be cleaned up, or wants a status check across all Ralph-loop/spike worktrees before running closeout on any of them.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 0 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use fleet-audit first, to see the whole picture.

    line 14
  2. Very specific:

    Use closeout second, worktree by worktree, only on things fleet-audit reported as safe-to-close — and still verify PR-merged status …

    line 15
  3. Very specific:

    Never use fleet-audit's output as authorization to force-remove anything.

    line 16
  4. Very specific:

    If that script or the private workspace state root (~/Development/encero/.agents-state/) doesn't exist on this machine, say so plainly and …

    line 35
  5. Very specific:

    If asked to mark something as protected on the spot, write a one-line reason to <worktree>/.agents/state/KEEP (create .agents/state/ first …

    line 40

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Fleet Audit — Incan Worktrees

Purpose

/fleet-audit gives a read-only, fleet-wide status report across every worktree under the shared tmp/ root — something closeout deliberately does not do, since closeout only acts on one PR/branch at a time. This skill never removes, moves, or modifies anything. It exists to tell you (or an orchestrator deciding what to hand to closeout) what's actually going on before anyone touches a worktree.

When to use this vs closeout

  • Use fleet-audit first, to see the whole picture.
  • Use closeout second, worktree by worktree, only on things fleet-audit reported as safe-to-close — and still verify PR-merged status yourself per closeout's own rules before removing anything.
  • Never use fleet-audit's output as authorization to force-remove anything. It classifies; it does not approve deletion.

How it works

Runs ~/Development/encero/.agents-state/scripts/fleet_audit.py, which:

  1. Lists every worktree via git worktree list --porcelain.
  2. Skips (reports as protected) any worktree with a .agents/state/KEEP marker file — that marker means a human already decided this one is not up for evaluation, full stop.
  3. For everything else, checks: branch ancestry against origin/main, matching GitHub PR state via gh pr list (catches squash/rebase merges the ancestor check alone would miss), uncommitted tracked file edits specifically (not just untracked build junk), and Ralph-loop slice status via the same parsing Codex's session hook already uses (~/.codex/hooks/ralph_state_impl.py's summarize_state).
  4. Classifies each as protected, safe-to-close, or needs-review, with a one-line reason.

Workflow

  1. Run:

    python3 ~/Development/encero/.agents-state/scripts/fleet_audit.py --repo-root . --repo-slug encero-systems/incan

    If that script or the private workspace state root (~/Development/encero/.agents-state/) doesn't exist on this machine, say so plainly and stop — do not attempt to reimplement the logic inline.

  2. Present the report grouped by status, exactly as the script outputs it. Do not editorialize the protected group — those are closed questions.
  3. For safe-to-close entries, note that the next step is a human-approved closeout run per worktree, not automatic removal.
  4. For needs-review entries, summarize the reason field per worktree so the user can decide quickly (tracked edits vs. non-terminal loop state vs. no signal at all) rather than re-reading the whole report.
  5. If asked to mark something as protected on the spot, write a one-line reason to <worktree>/.agents/state/KEEP (create .agents/state/ first if it doesn't exist) — do not mark something protected without a reason.

Output format

## Fleet Audit — <date>

### Protected (N)
- <worktree>: <reason>

### Safe to close (N)
- <worktree> (branch): <reason> — run /closeout to remove

### Needs review (N)
- <worktree> (branch): <reason>

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.