ck:code-review

Detect security flaws and failure modes in PR diffs and codebases via adversarial red-teaming review.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/jangtrinh/demo-Lam --skill ck-code-review-jangtrinh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ck:code-review
Source: https://github.com/jangtrinh/demo-Lam/tree/main/.claude/skills/code-review
Command: npx skills add https://github.com/jangtrinh/demo-Lam --skill ck-code-review-jangtrinh

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill eliminates risky, superficial reviews by running a three-stage, evidence-first code review process that includes an always-on adversarial red-team pass to uncover security holes, false assumptions, and real failure modes.

Core Features & Use Cases

  • Input-mode targeting (PR/commit/pending/codebase): Review the exact diff you care about (or scan the whole codebase) to keep scope accurate and avoid guesswork.
  • Three-stage rigor with gates: Enforce spec compliance first, then code quality, then adversarial review with a scope gate to avoid trivial noise while still attacking anything security-relevant.
  • Verification-first completion: Prevent “done” claims by requiring verification evidence before status, commit, or merge.

Quick Start

Run a pull request review by saying: review pull request #123 with full three-stage pipeline.

Frequently Asked Questions about ck:code-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run an adversarial security review on a pull request diff?

Run an adversarial security review on a pull request diff by targeting the specific PR to execute a three-stage pipeline enforcing spec compliance, code quality, and red-team failure mode analysis.

What is the best way to verify spec compliance before merging code changes?

The best way to verify spec compliance before merging is using an evidence-first code review process that gates code quality and adversarial security stages, preventing completion until verification evidence is provided.

Can I scan my whole codebase for security flaws and false assumptions?

Yes, you can scan your whole codebase for security flaws by running full-repo audits with optional parallel multi-reviewer coverage to manage larger scopes and uncover hidden false assumptions.

Does this code review process work with pending local changes?

Yes, the code review process works with pending local changes by resolving input-mode targeting to review exact local diffs, keeping scope accurate and preventing guesswork before commit.

How does adversarial red-teaming prevent superficial code reviews?

Adversarial red-teaming prevents superficial code reviews by applying an always-on security stage with scope gating to attack vulnerabilities and failure modes, eliminating risky guesswork through evidence-based detection.

When should I use parallel multi-reviewer coverage for commit diffs?

Use parallel multi-reviewer coverage for commit diffs when managing larger scopes during full-codebase audits, ensuring comprehensive adversarial analysis without generating trivial noise from scope gating.