reviewing
Performs formal reviews of code, architecture, and docs using established checklists.
Install
mkdir -p .claude/skills/reviewing && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/10289" && unzip -o skill.zip -d .claude/skills/reviewing && rm skill.zipInstalls to .claude/skills/reviewing
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.
The skill of performing thorough reviews of code, documentation, and releases to ensure quality, security, and readiness.Key capabilities
- →Review code
- →Review documentation
- →Review architecture
- →Verify release readiness
How it works
It applies canonical checklists to review files and records findings in specific markdown files.
Inputs & outputs
When to use reviewing
- →Review code for security vulnerabilities
- →Perform a formal documentation review
- →Check architecture against design models
- →Verify release readiness
About this skill
Reviewing Skill
General guidelines
- If a
docs/design/aurora/folder exists, read and follow Aurora Compact Model to understand the design.- The Aurora Compact Model exists to save time and tokens by keeping key information in a single compact (thus the name) file.
- Use the full Aurora instructions only when applying the Architecture skill.
- In order to ensure enough context for the review, break it down into manageable chunks and/or delegate it to subagents as needed. You may also ask for additional information or clarification from the user if necessary.
- If the user asks for a "formal review", treat it as a formal review and write the results to the appropriate files. Otherwise, provide recommendations in the conversation without writing to the review files.
- If unsure whether it is formal or not, ask.
- You MUST NOT alter code, documentation, or other files except the review output files.
- Reviews recommend improvements and fixes. Reviews do not make decisions and do not record decisions.
Review principles
- If reviewing code, refer to the Code Review Checklists and Principles for comprehensive guidance on what to look for. This includes security review guidance.
- If reviewing documentation, refer to the Documentation Review Checklists and Principles for comprehensive guidance on what to look for.
- If reviewing architecture and designs, refer to the Architecture Review Checklists and Principles for comprehensive guidance on what to look for.
- For a pre-release review, refer to the Pre-release Review Checklists for comprehensive guidance on what to look for.
- The checklist documents are canonical. If a checklist and a skill instruction conflict, the checklist wins.
Files and folders to ignore
- Read only the
Compact.jsonfiles underdocs/design/aurora/for context on the Aurora model. Do not read any other files in that directory for reviews. - Do not read any files or folders under
docs/design/namedMIS-*. Those are just the human-readable version of the model and contain no relevant information not already in the Compact Model. - Do not read files under
.github/unless specifically instructed to review them. - The following files are irrelevant for reviews:
.gitignore.gitattributes*.lock.editorconfigrustfmt.tomlrust-toolchain.tomlassets/
Deliverables
- Record findings in the appropriate review file under
docs/design/:docs/design/Review-Architecture.mddocs/design/Review-Code.mddocs/design/Review-Documentation.mddocs/design/Review-Prerelease.mddocs/design/Review-Security.md
- Apply the canonical checklist for the review type and record findings according to checklist requirements.
Cross-skill tasks
- If asked to implement fixes, do not proceed under this skill. Either:
- Switch to the appropriate skill, or
- Provide recommendations only and leave implementation to a separate task.
- If asked to update a project plan, use the Planning skill.
- If asked to produce or update an Aurora architecture model, use the Architecture skill.
Validation
- Findings are recorded in the correct file(s) under
docs/design/as specified above. - Findings are grouped by severity and include clear verification guidance.
- Recommendations are concrete and smallest-safe.
- The review output does not include implementation changes outside the review files.
Glossary
See the shared Skills glossary.
When not to use it
- →Implementing code fixes
- →Updating project plans
Prerequisites
Limitations
- →Cannot modify code or documentation
- →Must ignore specific irrelevant files
How it compares
It provides formal, checklist-driven reviews rather than informal feedback.
Compared to similar skills
reviewing side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| reviewing (this skill) | 0 | 5mo | No flags | Intermediate |
| cookbook-audit | 5 | 5mo | Review | Intermediate |
| code-review-checklist | 3 | 27d | No flags | Beginner |
| pr-review | 0 | 1mo | No flags | Advanced |
Try saying
Example prompts that trigger this skill in your AI assistant.
You might also like
cookbook-audit
anthropics
Audit an Anthropic Cookbook notebook based on a rubric. Use whenever a notebook review or audit is requested.
code-review-checklist
vudovn
Code review guidelines covering code quality, security, and best practices.
pr-review
microsoft
Multi-dimensional review of a PR or feature branch in the microsoft/winappcli repo. Activate when a contributor asks to "review my PR", "review my changes", "vet my branch before pushing", "do a full review", "PR review", "review this feature", or similar. Fans out parallel sub-agents covering secur
full-audit
oscardlfr
Run a unified audit across all quality dimensions — architecture, code quality, testing, security, docs, and release readiness. Consolidates findings from multiple agents and scripts into a single deduplicated report.
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.
github-code-review
ruvnet
Comprehensive GitHub code review with AI-powered swarm coordination