.claude/skills/review-feature/SKILL.md
Smart review that detects file types and launches the right review agents. Use when the user asks to review a feature, review changes, or after implementing a feature.
npx skillsauth add JLighter/dotfiles review-featureInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
You are a review coordinator. Your job is to detect what changed and launch the right review agents in parallel.
Run git diff --name-only HEAD to find modified files. If no diff, use the files or scope provided by the user.
Classify files into categories:
List the files and their categories.
Based on file categories, determine which reviews to run:
| Files detected | Agents to launch | |---------------|-----------------| | Domain/backend code | code-review + ddd-review | | Frontend components | code-review + ux-review | | Style files | css-review | | Frontend + styles | code-review + ux-review + css-review | | Full-stack | code-review + ddd-review + ux-review + css-review | | IaC files | infra-review | | Pipeline files | cicd-review | | IaC + pipelines | infra-review + cicd-review | | Documentation | pm-review (audit mode) |
code-review runs on ALL code changes (backend and frontend).
Announce which agents will be launched and why.
Launch all selected review agents simultaneously in a single message, each with the relevant file list.
IMPORTANT: Launch all selected agents in ONE message with multiple Agent tool calls. Do NOT launch them sequentially.
Once all agents return, produce a single unified report:
Date: current date Files reviewed: list Agents used: list
All critical findings from all agents, numbered, with:
All warnings, same format.
| Agent | Critical | Warnings | |-------|----------|----------| | Per agent | count | count | | Total | count | count |
The five most impactful things to fix, across all agents, prioritized by user and system impact.
$ARGUMENTS
development
Launch UX review (visual hierarchy, interaction, user flow). Use when the user asks to review UX, check UI, or after writing frontend components.
development
Deep security audit of the codebase. Traces data flows, validates findings adversarially, and proposes patches. Use for dedicated security audits, pen-test preparation, or when the user asks to scan for vulnerabilities.
development
Audit product documentation coherence against the codebase. Use when the user asks to check docs, verify documentation, or ensure docs are up to date.
development
Launch ALL four review agents in parallel (code, DDD, UX, CSS). Use when the user wants a comprehensive full review of everything.