agentleFS
Sign inStart free
System-Error-WorldwideSystem-Error-Worldwide/the-loop/.agents/skills/the-loop-doctor/SKILL.md

the-loop-doctor

Skill0 starsChanged 3 months agoMITView on GitHub

Inspect THE LOOP harness installation, skill discovery, collisions, permissions, configuration, and optional behavior evidence without changing local state. Use when a user asks whether the pack is installed, discoverable, compatible, shadowed, safe to run, or why a harness cannot find it.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check each harness executable and available version string independently.

    line 25
  2. Very specific:

    Run no behavior probe unless a library caller supplied an approved probe.

    line 29
  3. Distinctive:

    Read-only filesystem snapshot and deterministic performance tests.

    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

THE LOOP Doctor

Produce a read-only compatibility report for Codex, Claude Code, Kimi Code, and OpenCode. Discovery and behavior are separate claims.

Inputs

  • Repository root containing adapters/<harness>/adapter.json: the installed .the-loop/toolkit or the public checkout.
  • Project root to inspect.
  • Optional user home plus explicit CODEX_HOME and KIMI_CODE_HOME overrides for user-level roots.
  • Optional behavior-probe callback explicitly permitted by the caller. A verified result requires typed, matching harness/version/scope/capability/permission/environment evidence.

Procedure

  1. Validate every generic adapter manifest.
  2. Check each harness executable and available version string independently.
  3. Inspect adapter-declared project roots and, when supplied, user roots without following symlinked roots.
  4. Validate portable SKILL.md frontmatter, the complete 12-package identity, every source, duplicate name, and winning precedence. Call the pack complete only when the winning package digests and complete offline-toolkit digest match one unchanged Setup receipt.
  5. Report discovery as verified, failed, denied, or unverified. Keep installed: false separate.
  6. Run no behavior probe unless a library caller supplied an approved probe. Accept verified only for the exact portable-skill-invocation capability with the complete typed contract and the Doctor-derived harness, adapter, pack, and runtime environment digest; keep missing or mismatched proof non-ready.
  7. Inspect runtime version, config validity and permissions, state namespace safety, and kill-switch visibility without writing.

Core execution does not require network access or the source checkout. The CLI intentionally performs discovery only; live behavior probes belong to the separately approved conformance run.

From an installed project, use .the-loop/toolkit/scripts/the_loop_doctor.py. The canonical public source is scripts/the_loop_doctor.py, and the harness capability map defines status semantics.

Evidence

  • Overall status plus one report for each of the four harnesses.
  • Roots inspected, receipt-bound complete-pack status and digest, missing skills, collisions, issues, discovery status, typed behavior evidence, environment digest, and check time.
  • Read-only filesystem snapshot and deterministic performance tests.

Halt conditions

Label missing and invalid adapters, absent harnesses, denied roots, invalid packages, collisions, failed probes, and unverified probes precisely. Do not repair, overwrite, install, elevate permissions, invoke an undocumented command, or treat discovery as behavior proof.

Doctor does not collect prompts, send telemetry, or change the inspected project.

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.

Check your setup, free