agentleFS
y-miyazakiy-miyazaki/config/.agents/skills/agent-skills-review/SKILL.md

agent-skills-review

Skill1 starsChanged 4 days agoApache-2.0View on GitHub

Review SKILL.md quality for Waza readiness and agent-skills-instructions compliance. Use when evaluating SKILL.md quality, checking compliance with skill authoring standards, preparing skills for release, reviewing skill PRs, or running waza readiness checks.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 1 is distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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:

    Run validation scripts and parse results to assess SKILL.md quality against S-_, Q-_, P-_, and BP-_ checks.

    line 27
  2. Very specific:

    Check waza check token output and apply Q-09 per Workflow step 2.

    line 28
  3. Very specific:

    run benchmark content evaluations (waza run eval.yaml) as a substitute for compliance checks

    line 40
  4. Very specific:

    If target SKILL.md does not exist, set Status: failed per references/common-output-format.md and stop without running other checks.

    line 57
  5. Distinctive:

    If one validation script fails and the other succeeds, report successful checks normally and mark unresolved checks as deferred with script …

    line 61
  6. Very specific:

    If both validation scripts fail, set Status: failed and include both commands, exit statuses, and stderr summaries in ## Issues.

    line 62

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Input

  • Target (required): <agent-root>/skills/*/SKILL.md
  • Sibling skills (optional): other <agent-root>/skills/*/SKILL.md in the same package or sharing the target's contract files — enables P-04; without them P-04 is deferred
  • Evidence (required when not pre-supplied): outputs from scripts/validate_waza.sh and scripts/validate.sh — run both in Workflow step 1 when absent from context

Output Specification

Return structured Markdown in accordance with references/common-output-format.md. That file is the source of truth for the output contract.

Skill-local addition: unlike other review skills, this skill runs validation scripts, so Status: failed also applies when both scripts fail. The ## Issues cause entry (no ItemID) then lists both commands, exit statuses, and stderr summaries.

Execution Scope

  • Run validation scripts and parse results to assess SKILL.md quality against S-, Q-, P-, and BP- checks.
  • Check waza check token output and apply Q-09 per Workflow step 2.

USE FOR:

  • review new SKILL drafts before release
  • note token budget advisories when waza check exceeds 500 tokens (Q-09)
  • report SKILL spec compliance findings in PR reviews

DO NOT USE FOR:

  • implement product features
  • debug product runtime failures unrelated to SKILL quality checks
  • run benchmark content evaluations (waza run eval.yaml) as a substitute for compliance checks
  • modify or overwrite the target SKILL.md without explicit user approval (review output only; do not write or commit changes)
  • merge PRs or edit unrelated files
  • review product runtime behavior or application business logic

Reference Files Guide

  • common-checklist.md (always read)
  • common-output-format.md (always read)
  • category-best-practices.md (always read)
  • category-patterns.md (always read)
  • category-quality.md (always read)
  • category-structure.md (always read)
  • common-troubleshooting.md (read on failure)

Workflow

  1. If target SKILL.md does not exist, set Status: failed per references/common-output-format.md and stop without running other checks. Otherwise run bash scripts/validate_waza.sh <skill-name> and bash scripts/validate.sh <SKILL.md> (CWD: <agent-root>/skills/agent-skills-review/).
  2. Record token budget from waza check when present; if count > 500, add a Q-09 advisory entry to ## Issues and do not list Q-09 in ## Checks (Failed/Deferred Only). Q-09 fails only when waza check ran but its token count is not recorded; if validate_waza.sh failed, defer Q-09 per step 5. A skill compressed below sibling depth is a BP-04 finding, not Q-09.
  3. Apply checks in order: BP-* (best practices), P-* (workflow/policy), Q-* (quality language), S-* (structure). For Q-12, read every references/ file and the option parsing of bundled scripts/; for P-04, compare shared rules with the sibling skills.
  4. Report failed/deferred items with ItemIDs.
  5. If one validation script fails and the other succeeds, report successful checks normally and mark unresolved checks as deferred with script name and exit status.
  6. If both validation scripts fail, set Status: failed and include both commands, exit statuses, and stderr summaries in ## Issues.

Error Handling

ConditionSeverityAction
Target SKILL.md does not existFatalSet Status: failed; stop without other checks
validate_waza.sh or validate.sh missingFatalStop; report missing script path
One validation script fails, the other succeedsRecoverableReport successful checks; defer failed script checks with exit status
Both validation scripts failFatalSet Status: failed with command, exit status, and stderr summary in ## Issues
common-checklist.md unavailableFatalStop; report missing dependency
common-output-format.md unavailableRecoverableNote missing file; emit ## Checks Summary, ## Checks (Failed/Deferred Only), and ## Issues only
Script output missing after one rerunRecoverableDefer affected checks; include command and stderr per troubleshooting

Examples

  • Prompt: Review SKILL.md and report only failed/deferred items
  • Result: Structured report per references/common-output-format.md (failed/deferred only).

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.