一键导入
qa-review
Perform a thorough code review. Use when the user requests a review, wants to verify code quality, or before merging a PR.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Perform a thorough code review. Use when the user requests a review, wants to verify code quality, or before merging a PR.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Multi-agent team orchestration with native Agent Teams. Trigger when the user wants to launch a team of agents, coordinate parallel work with inter-agent communication, or use swarm mode.
Context transfer between AI sessions. Trigger when the user wants to save the context, resume a task, or hand off the work to another session.
API mock configuration for tests. Trigger when the user wants to mock APIs, use MSW, or test without a backend.
Debug and resolve problems. Use when the user has a bug, an error, an unexpected behavior, or wants to understand why something is not working.
TDD development with Red-Green-Refactor cycle. Use to implement a feature by writing tests BEFORE the code. Trigger automatically when the user asks for TDD, wants to write tests first, mentions "test first", or asks to implement, add, create, fix, correct code, a new feature, a bugfix, or a functionality.
Orchestration of parallel agents to maximize efficiency. Trigger when a task can be decomposed into independent sub-tasks that can run in parallel.
| name | qa-review |
| description | Perform a thorough code review. Use when the user requests a review, wants to verify code quality, or before merging a PR. |
| allowed-tools | ["Read","Glob","Grep"] |
| context | fork |
Identify quality, security, and maintainability issues BEFORE merge.
Claude Code ships a native /code-review that owns the review execution: effort levels from low to max, --fix to apply findings, --comment to post inline PR comments, and ultra for a multi-agent cloud review. Prefer it to run the review.
This skill's delta is the conventions the review is held to — the checklist below (incl. the substance-check.sh gate native review does not run), the naming tables, and the severity taxonomy. Use them to brief or grade a native /code-review run, or as the manual protocol where the native command is unavailable.
# View the changes
git diff main...HEAD --stat
git log main...HEAD --oneline
any./scripts/substance-check.sh <changed-files> (flags no-assertion / always-true / skipped / empty / stub; a green suite over hollow tests is not "done")[TYPE] file:line - comment
Types:
- [CRITICAL] - Blocking, must be fixed
- [IMPORTANT] - Should be fixed
- [SUGGESTION] - Optional improvement
- [QUESTION] - Clarification needed
- [NITPICK] - Minor detail
## Review: [PR Title]
### Summary
- **Files modified**: X
- **Lines added**: +Y
- **Lines removed**: -Z
- **Verdict**: Approve / Request Changes / Comment
### Positive points
- [Point 1]
- [Point 2]
### Issues identified
#### Critical
- [CRITICAL] `file.ts:42` - Description
#### Important
- [IMPORTANT] `file.ts:87` - Description
### Suggestions
- [SUGGESTION] `file.ts:123` - Description
### Final checklist
- [ ] Code readable and maintainable
- [ ] Sufficient tests
- [ ] No security issue
- [ ] Acceptable performance
| Element | Convention | Good examples | Bad examples |
|---|---|---|---|
| Variables | Descriptive, camelCase | userCount, isActive | x, tmp, data |
| Functions | Verb + noun, camelCase | getUserById, validateEmail | process, handle, do |
| Booleans | Prefix is/has/can/should | isValid, hasPermission | valid, permission |
| Constants | SCREAMING_SNAKE | MAX_RETRY_COUNT | maxRetry |
| Classes | PascalCase, noun | UserService, OrderRepository | Manager, Helper |
| Interfaces | PascalCase, descriptive | UserProfile, PaymentMethod | IUser, DataType |
| Smell | Problem | Fix |
|---|---|---|
| Generic name | data, result, temp, info | Name based on content |
| Abbreviation | usr, btn, msg, idx | Write in full |
| Double negation | !isNotValid, !disableButton | isValid, enableButton |
| Type in the name | userArray, nameString | users, name |
| Inappropriate length | Short global variable, long local | Reverse: long global, short local |
| Misleading name | getUser that modifies | fetchAndUpdateUser |
# Single-character variables (except i, j in loops)
\b[a-z]\b\s*[=:]
# Generic names
\b(data|result|temp|tmp|info|item|obj|val|res)\b\s*[=:]
# Booleans without prefix
\b(active|valid|visible|enabled|disabled|open|closed)\b\s*[=:]
The formerly-recommended official code-review plugin is superseded: /code-review is now native in Claude Code at multiple effort levels (incl. the multi-agent cloud ultra tier) — no plugin install needed. This skill keeps the checklist + conventions; the native command owns the orchestration.
Full list of validated vendor skills: docs/recipes/recommended-vendor-skills.md. Audit pilot trace: specs/marketplace-audit/qa-skills-pilot-2026-05-06.md.