Skip to content
AI & Agents
Skill

/code-review

Review diffs and change sets for bugs, regressions, risks, and missing tests.

BOOST
From plugin
hybridclaw
15989 skills
Install
$ npx -y skills add HybridAIOne/hybridclaw --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.

Review diffs and change sets for bugs, regressions, risks, and missing tests.

SKILL.md

code-review.SKILL.md
name: code-review
description: Review diffs and change sets for bugs, regressions, risks, and missing tests.
user-invocable: true
metadata:
  hybridclaw:
    category: development
    short_description: "Diff and regression review."
    tags:
      - engineering
      - review
      - quality
      - github
    related_skills:
      - github-pr-workflow
      - feature-planning

Code Review

Use this skill to review local changes or an existing GitHub pull request for correctness, regressions, maintainability, security, and missing tests.

Review Order

1. Establish scope with `git status --short`, `git diff --stat`, and `git diff --name-only`. 2. Read the changed files, not just the diff, before judging behavior. 3. Run targeted validation when it exists: tests, typecheck, lint, build, or repo-specific checks. 4. Present findings ordered by severity, then open questions, then a brief summary.

Primary Workflows

Local Branch Review

Use local git state when the change is in the current checkout:

git status --short
git diff --stat
git diff --name-only
git diff --staged
git diff <base-branch>...HEAD
git log --oneline <base-branch>..HEAD

Replace `<base-branch>` with the repository's actual review base branch.

GitHub PR Review

Use `gh` when the review target is an open pull request:

gh pr view 123
gh pr diff 123
gh pr checkout 123
gh pr view 123 --comments

After checking out the PR branch, review it the same way as a local branch and run the relevant repo checks before leaving comments.

What to Look For

Prioritize issues that change behavior or raise delivery risk:

  • incorrect logic or broken edge cases
  • state, data, or migration regressions
  • auth, permission, or secret-handling mistakes
  • missing validation, retries, or error handling
  • flaky or incomplete tests
  • risky coupling, hidden side effects, or cleanup gaps

Treat pure style comments as low priority unless the user explicitly asks for a style review.

Review Output

Default to this structure:

1. Findings 2. Open questions or assumptions 3. Brief summary

For each finding:

  • cite the file and the most relevant line or function
  • explain the concrete failure mode or risk
  • describe the user-visible impact when possible
  • note the missing test or validation that would catch it

If there are no findings, say that explicitly and mention any residual risk such as unrun integration tests or unverified deployment paths.

Working Rules

  • Prefer evidence from code and test behavior over speculation.
  • Distinguish confirmed defects from probable risks and from optional

suggestions.

  • Review generated or vendored files by tracing back to their source inputs when

practical.

  • Do not bury the most severe issue behind minor nits.
  • When a diff is large, identify the riskiest files first and focus depth there.

Useful Checks

Use targeted searches when helpful:

git diff <base-branch>...HEAD | rg "console\\.log|TODO|FIXME|HACK|debugger"
git diff <base-branch>...HEAD | rg "<<<<<<<|=======|>>>>>>>"
git diff <base-branch>...HEAD | rg -i "password|secret|token|api[_-]?key|private[_-]?key"

Prefer repo-native test commands over broad guesses. Run the smallest check set that can confirm or disprove a suspected issue.

Read more
Ships withhybridclaw

Enterprise-ready self-hosted AI assistant runtime with sandboxed execution, secure credentials, approvals, and memory

Get the whole plugin
Stats
159
Stars
17
Forks
Active
Maintenance
TypeScript
Language
MIT
License
13h ago
Last commit
8mo ago
Created
3h ago
Added

Repo: HybridAIOne/hybridclaw

Other skills on hybridclaw.