Skip to content
Agent Orchestration
Skill

/code-review

Use when the user asks for a code review of local files, a workspace, a branch diff, or a pull request. Focus on real bugs, regressions, accessibility issues, and misleading behavior, with findings ordered by severity.

BOOST
From plugin
flowpilot
13620 skills
Install
$ npx -y skills add 6BNBN/FlowPilot --skill code-review --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/code-review

Context preview

The summary Claude sees to decide when to auto-load this skill.

Use when the user asks for a code review of local files, a workspace, a branch diff, or a pull request. Focus on real bugs, regressions, accessibility issues, and misleading behavior, with findings ordered by severity.

SKILL.md

code-review.SKILL.md
name: code-review
description: Use when the user asks for a code review of local files, a workspace, a branch diff, or a pull request. Focus on real bugs, regressions, accessibility issues, and misleading behavior, with findings ordered by severity.

Code Review

Goal

Find real issues first. Prioritize correctness, regressions, accessibility, security, and misleading UX over style commentary.

Scope Selection

  • If the user names files or paths, review those first.
  • If there are local changes, review the changed files and relevant nearby context.
  • If a branch diff or pull request is provided, review the diff and then read surrounding code only where needed.
  • If scope is unclear, ask what should be reviewed instead of scanning the entire repo blindly.

Review Workflow

1. Understand the intent of the change. 2. Inspect the relevant files, diffs, and nearby code paths. 3. Look for real execution risks: wrong logic, missing edge cases, state bugs, API misuse, broken UX flows, accessibility problems, and misleading UI behavior. 4. Validate each suspected issue against the actual code before reporting it. 5. Report only high-signal findings.

Output Format

  • List findings first, ordered by severity.
  • For each finding include:
  • severity
  • file or scope
  • concise explanation
  • suggested fix when clear
  • If no findings are found, say so explicitly and mention residual risks or testing gaps.

Avoid

  • Praise-first reviews that bury the real issues.
  • Pure formatting or style nitpicks unless the user asked for them.
  • Reporting speculative issues without evidence.
  • Treating missing tests as the main finding unless the risk is concrete.

Optional PR Workflow

If GitHub or git context is available, include PR or diff context. If not, do a local review and clearly state what information was unavailable.

Read more
Ships withflowpilot

一个文件,一句开发需求,全自动开发。 把 flow.js 丢进任何项目,打开你选择的客户端(Claude Code / Codex / Cursor / snow-cli 等)描述你要做什么,然后去喝杯咖啡。 回来的时候,代码写好了,测试跑完了,git 也提交了。 新增说明:现已兼容 Claude Code、Codex、Cursor、snow-cli 和其他客户端;init 时可直接选择目标客户端并生成对应的 instruction file / 配置。 新增说明:内置 instruction file /

Get the whole plugin

Other skills on flowpilot.