code-review_skill

This skill enforces mandatory automated code reviews before commits and deploys, improving security, quality, and reliability across projects.

Shell

450

GitHub Stars

1

Bundled Files

3 weeks ago

Catalog Refreshed

2 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstart where the catalogue uses aiagentskills.

npx veilstart add skill alinaqi/claude-bootstrap --skill code-review

SKILL.md30.7 KB

Overview

This skill enforces mandatory AI-assisted code reviews via the /code-review command before commits, merges, and deployments. It supports Claude by default and optional engines like OpenAI Codex and Google Gemini, with dual or triple-engine modes for higher coverage. The workflow blocks commits on critical or high-severity findings and integrates with pre-commit hooks and CI/CD pipelines.

How this skill works

When invoked, the skill scans staged changes, files, or PR diffs and runs one or more review engines to analyze security, performance, architecture, tests, and docs. Multi-engine runs compare, deduplicate, and classify findings by severity and consensus (unanimous, majority, single). Results include actionable locations, severity levels, and a blocked/passed status used by hooks and CI steps.

When to use it

Before every commit to enforce a mandatory gate
On pull requests to surface issues before merge
As a final pre-deploy checklist for production releases
For security-critical code (auth, payments, secrets) use multi-engine mode
When onboarding or auditing large codebases to find structural and testing gaps

Best practices

Set a sensible default engine in user or project settings and override per-need
Use dual/triple engine modes for security-sensitive areas or critical PRs
Integrate the pre-commit hook so commits are blocked on Critical/High findings
Focus review runs with --focus flags (security, performance, architecture) for targeted checks
Tune CI pipeline to run fast single-engine checks on PRs and full multi-engine scans on release branches

Example use cases

Run /code-review on staged TypeScript changes via the pre-commit hook to prevent SQL injection and auth regressions
Use /code-review --pr 123 in CI to produce a review comment on a pull request before merge
Invoke /code-review --focus performance to detect N+1 queries and memory leaks in backend services
Run triple-engine mode for payment or SSO subsystems to maximize detection coverage
Scan large repo diffs with Gemini for long-context analysis and Codex for high-accuracy findings

FAQ

Yes. The included pre-commit hook and CI steps inspect the review output and exit non-zero if Critical or High issues are found, preventing commits or merges.

Which engine should I choose by default?

Use Claude for fast, no-setup reviews. Use Codex for higher accuracy in CI, Gemini for very large contexts, and dual/triple modes for security-critical code.

How are multi-engine findings combined?

Findings are deduplicated and classified by consensus: unanimous (all engines), majority (two of three), or single-engine. The summary shows counts and a blocked/passed status.

Can I target specific files or concerns?

Yes. Use file path arguments, PR IDs, or --focus flags (security, performance, architecture) to tailor reviews to relevant areas.