RE

review-quality

Audits repository quality and converts findings into fixable, owner-scoped tasks.

Install

mkdir -p .claude/skills/review-quality-lurek2d && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/12529" && unzip -o skill.zip -d .claude/skills/review-quality-lurek2d && rm skill.zip

Installs to .claude/skills/review-quality-lurek2d

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Load this skill when auditing overall repo quality, hotspots, contract drift, and tool-reported quality findings. Skip it for narrow module implementation tasks with clear requested edits.
188 chars✓ has a “when” trigger
Intermediate

Key capabilities

  • Audit overall quality signals
  • Convert tool findings into fixable work
  • Group findings by severity and owner subsystem
  • Verify tool findings against source
  • Address narrow high-confidence issues
  • Report changed files and validation evidence

How it works

The skill runs quality reports and audit tools, then groups findings by severity and owner subsystem. It verifies tool findings against the source before recommending or applying fixes.

Inputs & outputs

You give it
repository context and user request for quality audit
You get back
audit output with findings, or fixed issues with revalidation

When to use review-quality

  • Auditing repo quality
  • Identifying architectural hotspots
  • Fixing contract drift

About this skill

review-quality

Mission

  • Audit overall quality signals and convert tool findings into fixable owner-scoped work.

Domain Knowledge

  • quality_report.py aggregates repository quality signals for triage.
  • One root defect can appear in coverage, docs, wrapper, and contract reports.
  • Duplicate symptoms do not count as separate root risks.
  • File size and warning count are investigation hints, not defects by themselves.
  • A hotspot combines broad responsibility or churn with weak ownership, missing proof, unsafe input, or contract drift.
  • A large coordination module can be coherent.
  • A smaller helper can be a hotspot when it mixes parsing, mutation, I/O, and presentation.
  • A parser-enforced filename, marker, registry, or schema is operational contract truth.
  • The nearest AGENTS.md, skill, template, fixture, and test must match parser behavior.
  • Generated output is fixed through its canonical source or generator.
  • Vendored files are not normal repository source owners.
  • Test fixtures can contain intentional invalid or old shapes.
  • Compatibility shims can contain intentional duplication.
  • Provenance must be checked before proposing a fix.
  • A useful finding names the violated invariant, evidence, owner, consequence, and acceptance command.
  • “Refactor this file” is not an actionable finding.
  • Allowlists, suppressions, baselines, and compatibility exceptions can become stale.
  • An exception needs a current reason and a narrow matcher.
  • Total finding count can stay flat while severity and ownership get worse.
  • Quality comparison uses root-cause clusters and severity, not only totals.
  • cag_validate.py checks CAG contracts but does not validate product behavior.
  • The master quality dashboard combines docs-general, test coverage, module audit, and API validation payloads.
  • Its current documentation gates are 90% for Rust items and 50% for Lua API functions.
  • Its current test gates are 50% for Rust functions and 30% for Lua functions.
  • Any API validation issue prevents an overall PASS.
  • A child audit result with malformed JSON is reported as a tool error, not a product finding.
  • Quality report exit codes distinguish pass (0), failed gates (1), and fatal report failure (2).
  • Aggregate percentages do not replace exact owner audits for examples, tests, specs, or CAG routing.

Workflow

  1. Read root, source, and tools contracts for the requested scope.
  2. Run quality_report.py for that scope.
  3. Group raw findings by canonical source owner.
  4. Merge coverage, docs, wrapper, and contract symptoms with the same root cause.
  5. Exclude generated, vendored, fixture, and intentional compatibility symptoms only after checking provenance.
  6. Read the enforcing parser or validator for each contract finding.
  7. Compare parser behavior with the nearest guidance and tests.
  8. Inspect source and proof for each hotspot candidate.
  9. Score severity, confidence, breadth, and remediation cost.
  10. Keep narrow, high-confidence findings.
  11. Mark current-scope regressions separately from existing backlog.
  12. Record invariant, exact evidence, owner, consequence, and acceptance command.
  13. Inspect allowlists, suppressions, baselines, and exceptions in the same subsystem.
  14. Reject exceptions without a current reason or narrow matcher.
  15. Fix the canonical owner only when fixes are requested.
  16. Sync parser guidance, fixtures, and tests when the contract changed.
  17. Run the focused audit after each fix.
  18. Run cag_validate.py when CAG surfaces changed.
  19. Rerun quality_report.py.
  20. Compare root-cause clusters and severity before and after fixes.
  21. Inspect new findings to confirm responsibility was removed, not moved.

References

  • contracts: AGENTS.md, src/AGENTS.md, tools/AGENTS.md
  • tools: tools/python.cmd tools/rag/query.py "quality report hotspots contract drift" --profile all --limit 10, tools/python.cmd tools/audit/quality_report.py, tools/python.cmd tools/validate/cag_validate.py
  • agent: reviewer
  • RAG: quality report hotspots contract drift <scope>; inspect the aggregate finding, enforcing parser/contract, canonical source owner, tests, and active exceptions.

When not to use it

  • Narrow module implementation tasks with clear requested edits

Limitations

  • Cannot proceed if required user intent, target module, or validation threshold is missing
  • Cannot proceed if a referenced owner path or tool is absent
  • Cannot fix findings that require changing unrelated user work or widening scope

How it compares

This workflow prioritizes tool-based auditing and verification before manual inspection or fixes, unlike a manual approach that might start with broad file reads.

Compared to similar skills

review-quality side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
review-quality (this skill)02moNo flagsIntermediate
architect-review1094moNo flagsAdvanced
solid-principles579moNo flagsIntermediate
codex322moReviewAdvanced

Try saying

Example prompts that trigger this skill in your AI assistant.

You might also like

architect-review

sickn33

Master software architect specializing in modern architecture patterns, clean architecture, microservices, event-driven systems, and DDD. Reviews system designs and code changes for architectural integrity, scalability, and maintainability. Use PROACTIVELY for architectural decisions.

109320

solid-principles

SmidigStorm

Enforce SOLID principles (Single Responsibility, Open/Closed, Liskov Substitution, Interface Segregation, Dependency Inversion) in object-oriented design. Use when writing or reviewing classes and modules.

57236

codex

Lucklyric

Invoke Codex CLI for complex coding tasks requiring high reasoning capabilities. This skill should be invoked when users explicitly mention "Codex", request complex implementation challenges, advanced reasoning, or need high-reasoning model assistance. Automatically triggers on codex-related requests and supports session continuation for iterative development.

32238

error-handling-patterns

wshobson

Master error handling patterns across languages including exceptions, Result types, error propagation, and graceful degradation to build resilient applications. Use when implementing error handling, designing APIs, or improving application reliability.

35170

deepwiki-rs

sopaco

AI-powered Rust documentation generation engine for comprehensive codebase analysis, C4 architecture diagrams, and automated technical documentation. Use when Claude needs to analyze source code, understand software architecture, generate technical specs, or create professional documentation from any programming language.

25170

senior-fullstack

davila7

Comprehensive fullstack development skill for building complete web applications with React, Next.js, Node.js, GraphQL, and PostgreSQL. Includes project scaffolding, code quality analysis, architecture patterns, and complete tech stack guidance. Use when building new projects, analyzing code quality, implementing design patterns, or setting up development workflows.

35110

Search skills

Search the agent skills registry