Skip to main content
skill CX Codex

recipe-quality-profile — Claude Code skill: define code review policy

recipe-quality-profile is a skill that proposes repository-specific code quality policy and saves it to docs/project-context/quality.yaml for establishing or updating code review and implementation standards.

MIT ★ 720 updated 2026-10-29

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add shinpr/claude-code-workflows --skill recipe-quality-profile -a claude-code
Open on GitHub

What recipe-quality-profile does

The skill examines your repository — contributor docs, CI config, manifests, tests, and code patterns — and builds a set of review rules. Each rule states when it applies, what code state passes review, and why this repository owns that decision.

Before writing, it presents all proposed changes, conflicting evidence if found, and asks for confirmation. After you approve, it creates or updates the quality profile — a YAML file with version 1, unique rule IDs, and citations from your code.

It first runs two supporting skills: llm-friendly-context to capture repository context and coding-principles to separate repository-owned policy from general code-quality knowledge.

What it helps with

  • Write down code review standards tied to your repository's actual code
  • Update quality policy when requirements shift
  • Distinguish decisions your project owns from universal practices
  • Document why your repository follows specific patterns

Example prompts for the agent

  • Propose a quality profile for our project based on the current code and CI setup
  • Update quality.yaml to require type annotations in new code
  • Create a quality profile that separates our standards from general best practices

How to install recipe-quality-profile

1Skills CLI (Vercel)

npx skills add shinpr/claude-code-workflows --skill recipe-quality-profile -a claude-code

2Manually: put the files in place

~/.claude/skills/recipe-quality-profile/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/recipe-quality-profile/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about recipe-quality-profile

How do I install recipe-quality-profile in Claude Code?

Run npx skills add shinpr/claude-code-workflows --skill recipe-quality-profile -a claude-code, then read ~/.claude/skills/recipe-quality-profile/SKILL.md.

What does recipe-quality-profile do?

It analyzes your code, tests, and config files, proposes review rules specific to your repository, and saves them to docs/project-context/quality.yaml after your approval.

What other skills does recipe-quality-profile depend on?

The skill automatically invokes llm-friendly-context to build repository context and coding-principles to separate repository-specific rules from general practices.

Source text

SKILL.md · MIT
---
name: recipe-quality-profile
description: Proposes repository-specific quality policy for implementation and review and, after confirmation, creates or updates docs/project-context/quality.yaml. Use when asked to create or update a repository quality profile.
disable-model-invocation: true
---

Execute Skill: llm-friendly-context before proposing profile conditions or writing the profile.
Execute Skill: coding-principles to distinguish repository-owned policy from general code-quality knowledge.

## Purpose

Establish repository-specific implementation and code-review acceptance conditions with the user, then create or update `docs/project-context/quality.yaml`.

Requested policy change: $ARGUMENTS

## Profile Contract

```yaml
version: 1
review_dimensions:
  - id: stable-kebab-case-id
    applies_when: Observable condition that makes this repository rule relevant to a change.
    pass: Observable accepted state to verify.
    evidence:
      - "repository/path: section, identifier, or contract"
```

Each dimension owns one repository-specific quality decision. `applies_when` limits its implementation and review surface, `pass` defines the accepted state, and `evidence` identifies why the repository owns the rule.

## Authoring Flow

1. Use the current working repository as the target and read its `docs/project-context/quality.yaml` when present. For an existing profile, use the requested policy change as the update boundary; when none is supplied, ask for it and keep the profile unchanged.
2. Build candidates from acceptance conditions expressed or enforced by repository instructions, contributor documentation, CI, manifests and scripts, schemas and public contracts, tests, or representative implementation patterns. For an update, derive candidates only from the requested policy change and preserve unrelated dimensions.
3. For each candidate, inspect supporting and contradicting evidence wherever it can change the candidate's applicability, accepted state, or repository ownership. Separate observed repository facts from policy choices that require user confirmation.
4. Retain a candidate only when failing its `pass` condition would change implementation acceptance and every repository fact it depends on has cited evidence. Give it the narrowest useful `applies_when`, one positive observable `pass` condition, and consolidate candidates that would produce the same finding and correction. Omit a candidate when required repository evidence is unavailable and report the exact evidence needed.
5. Present proposed additions, changes, and removals, confirm that other dimensions remain unchanged, show the supporting and contradicting evidence, and state unresolved policy choices with their effect on implementation and review acceptance. Obtain explicit user confirmation of a proposal with no unresolved choices before writing.
6. Write only the confirmed profile content. Read the result and verify version `1`, unique IDs, all required fields, observable conditions, readable evidence references, and consistency with the confirmed proposal.

When no repository-specific dimension remains and no profile exists, report that repository evidence supports no profile content and leave the repository unchanged.

## Result

Before confirmation, report:

- proposed additions, changes, and removals, plus the unchanged remainder;
- supporting and contradicting evidence for each modification;
- omitted candidates and the exact missing evidence;
- policy choices requiring the user's decision and their effect on implementation and review acceptance.

After confirmation, report:

- the profile path and whether it was created, updated, or left unchanged;
- dimensions added, changed, or removed;
- evidence used for each changed dimension;
- an exact validation limitation when the result could not be verified.

## Completion Check

- [ ] Every retained dimension changes an implementation or review decision and cites repository or user-confirmed policy evidence
- [ ] Conditions are positive, observable, and limited by `applies_when`
- [ ] The profile contains repository-specific acceptance conditions only
- [ ] Supporting and contradicting evidence were compared where they could change the proposal
- [ ] The user confirmed the complete proposal before the repository write
- [ ] Dimensions outside the update boundary remain unchanged
- [ ] The written profile satisfies the Profile Contract

© shinpr · MIT · shinpr/claude-code-workflows

Similar skills

All skills for Code review and refactoring →
improve-code-quality wondelai improve-code-quality — a skill for Claude Code and Codex guides untested code toward production stability. It runs an interactive nine-phase workflow and records every milestone in local markdown files. CX Codex +1 2,700installs code-review-and-quality addyosmani code-review-and-quality — a skill for Claude Code and other platforms runs multi-axis code reviews across five quality dimensions. Runs before merging any pull request. CX Codex +1 92,300installs pr-quality-checklist bobmatnyc pr-quality-checklist — a skill for Claude Code, Codex, Cursor and other agents: runs a structured PR checklist before the pull request goes up for review. Invoked manually by the author, not triggered automatically. CX Codex +1 354installs quality-gates OrchestKit quality-gates scores task complexity on a 1–5 scale and halts work when requirements are vague or solutions loop. Triggers before complex tasks and after repeated failures. CX Codex +1 259installs profile-resolver serpro69 profile-resolver — a subagent that reads a diff and the worktree, works out which project profiles are active, and reports which checklists load. Handy when detection has to run apart from the review. CX Codex ★ 155repo stars thermo-nuclear-code-quality-review cursor thermo-nuclear-code-quality-review is a Claude Code skill for a strict review of maintainability and code structure. Ask for it to audit changes on the current branch and find ways to simplify them. CX Codex +1 24,700installs

From the same set43

The whole set →
ui-analyzer ui-analyzer collects decision-relevant UI evidence from the codebase and recorded external sources. Use it before drafting a UI Spec or Design Doc. CX Codex ★ 720repo stars external-resource-context external-resource-context tracks access details for external API schemas, design files, IaC, and secret stores outside the repository. Prevents repeated discovery of environment assets. CX Codex +1 ★ 720repo stars task-decomposer task-decomposer turns each task in an approved Work Plan into a separate implementation file. Use it when the plan is ready to hand off for execution. CX Codex ★ 720repo stars acceptance-test-generator acceptance-test-generator — a subagent for Claude Code generates integration and E2E test skeletons from Design Doc acceptance criteria. Triggers when planning test suites or on requests mentioning test skeletons. CX Codex ★ 720repo stars requirement-analyzer requirement-analyzer inspects the codebase to collect scope and cost evidence before confirming new implementation requirements. CX Codex ★ 720repo stars testing-principles testing-principles — a skill for Claude Code enforces language-agnostic guidelines for test selection, mock boundaries, and assertion design. Triggers when writing new tests, planning test scope, or reviewing test suites. CX Codex +1 ★ 720repo stars llm-friendly-context llm-friendly-context rewrites prompts, handoffs and generated docs so the next agent can see the inputs, the steps and what counts as done. Useful whenever text goes to another model. CX Codex +1 ★ 720repo stars requirement-convergence requirement-convergence — a Claude Code skill that isolates a project outcome from proposed requirements and records user-stated exclusions. Runs at requirement intake before any system design begins. CX Codex +1 ★ 720repo stars

Project sponsors

Companies that keep this analytics open