agentleFS
y-miyazakiy-miyazaki/config/.claude/skills/instructions-review/SKILL.md

instructions-review

Skill1 starsChanged 3 days agoApache-2.0View on GitHub

Review instruction and rule files for structure, consistency, applyTo precision, and portable cross-references — package sources (`*.instructions.md`) and distributed agent rule targets (`.cursor/rules/*.mdc`, `.claude/rules/*.md`, `.kiro/steering/*.md`). Use when reviewing instruction PRs, audits, or distributed rule changes after package sync.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 8 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    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:

    Do not modify instructions files or approve/merge PRs

    line 40
  2. Distinctive:

    Keep chapter order and section naming aligned with sibling instruction/rule files in the same repository (source or distributed form).

    line 41
  3. Distinctive:

    create new instruction files from scratch

    line 52
  4. Distinctive:

    run syntax/security validation tooling

    line 53
  5. Distinctive:

    Read PR context and identify target instruction files.

    line 70
  6. Very specific:

    If a target file does not exist, mark it missing in ## Targets (not a check failure, so no ## Issues entry) and continue remaining targets.

    line 71
  7. Distinctive:

    If PR context is unavailable, identify targets from the user prompt and review the files alone.

    line 72
  8. Distinctive:

    Verify required chapter order (Scope → Standards → Guidelines → Testing and Validation → Security Guidelines), then review checklist …

    line 73
  9. Distinctive:

    No empty sections: every H3 must have content

    line 89

≈ “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 instruction/rule files (required): package sources (**/instructions/*.instructions.md) and/or distributed targets (.cursor/rules/*.mdc, .claude/rules/*.md, .kiro/steering/*.md)
  • PR context (optional)

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: insert a ## Targets section directly after the # <Result Title> heading, before ## Checks Summary, with one row per target file:

## Targets

| File     | Status                |
| -------- | --------------------- |
| `<path>` | `reviewed`, `missing` |

Then, in ## Issues and ## Checks (Failed/Deferred Only), prefix each entry with the file path when more than one target is reviewed. Give each Deferred entry its reason.

Execution Scope

  • Systematically apply review checklist from references/common-checklist.md
  • Focus on quality, structure, consistency, and practical usability requiring human/AI judgment
  • Do not execute validation commands from this review skill
  • Do not modify instructions files or approve/merge PRs
  • Keep chapter order and section naming aligned with sibling instruction/rule files in the same repository (source or distributed form).

USE FOR:

  • review package *.instructions.md and distributed Cursor/Claude/Kiro rule files in PRs
  • audit applyTo precision, stem-based cross-references, and companion coverage (G-03, G-04, G-05)
  • audit instruction quality and rule consistency
  • propose review findings without modifying files

DO NOT USE FOR:

  • create new instruction files from scratch
  • run syntax/security validation tooling
  • review general markdown docs outside instruction/rule paths (e.g., README.md, docs/*.md, skills)

Reference Files Guide

  • common-checklist.md (always read)
  • common-output-format.md (always read)
  • common-troubleshooting.md (read when a check fails)
  • category-global.md (always read)
  • category-testing.md (always read)
  • category-security.md (always read)
  • category-quality.md (always read)
  • category-guidelines.md (always read)
  • category-standards.md (always read)

Workflow

  1. Read PR context and identify target instruction files.
  2. If a target file does not exist, mark it missing in ## Targets (not a check failure, so no ## Issues entry) and continue remaining targets. Process all target files independently — one failure does not stop the review of others.
  3. If PR context is unavailable, identify targets from the user prompt and review the files alone. No checklist item requires PR context, so nothing is deferred for its absence.
  4. Verify required chapter order (Scope → Standards → Guidelines → Testing and Validation → Security Guidelines), then review checklist priorities and collect failed/deferred ItemIDs.
  5. Output required report sections per references/common-output-format.md.

Severity and Status Rules

StatusWhen to use
FailedCheck evaluated and issue confirmed with concrete evidence (file + line or missing element)
DeferredCheck cannot be evaluated — its category file is unavailable
PassedCheck evaluated and no issue found (counted in summary only)

Key evaluation criteria (inline summary of common-checklist):

  • Structure: 5 H2 chapters in correct order (Scope → Standards → Guidelines → Testing and Validation → Security Guidelines)
  • Guidelines format: H3 headings with category IDs; rule bullets are ItemID + (LEVEL) + title only (synced instructions are thin — do not require Check: child bullets in ## Guidelines; full Check/Why/Fix live in *-review category-*.md)
  • Code Modification Guidelines: required as the last Guidelines subsection (STRUCT-05, MUST); the clarity of its content is GUIDE-01 (SHOULD)
  • No empty sections: every H3 must have content

Error Handling

ConditionSeverityAction
common-checklist.md unavailableFatalStop, report missing dependency
common-output-format.md unavailableRecoverableNote missing file; emit ## Checks Summary, ## Checks (Failed/Deferred Only), and ## Issues only
Category reference file missingRecoverableDefer checks from that category, note missing file in report
All category files missingFatalStop, report "no evaluation criteria available"
Target instruction/rule file does not existRecoverableMark the file missing in ## Targets, continue remaining targets
All target files missingFatalSet Status: failed; stop; ## Issues cause entry: "no reviewable instruction files found"

Examples

  • Prompt: Review instruction and rule files for structure and portable cross-references
  • Result: Structured report per references/common-output-format.md (per-file failed/deferred ItemIDs).

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.