bmad-code-review

Review code changes adversarially and triage findings into actionable outcomes.

Updated Apr 23, 2026
One-click install
npx skills add https://github.com/ncptuan/stock_backtest_project --skill bmad-code-review-ncptuan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: bmad-code-review
Source: https://github.com/ncptuan/stock_backtest_project/tree/main/.claude/skills/bmad-code-review
Command: npx skills add https://github.com/ncptuan/stock_backtest_project --skill bmad-code-review-ncptuan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you review code changes rigorously when a normal pass might miss hidden defects, spec mismatches, or risky edge cases. It structures review into multiple adversarial perspectives so findings are more reliable and easier to act on.

Core Features & Use Cases

  • Layered Review Strategy: Runs parallel review roles including Blind Hunter, Edge Case Hunter, and Acceptance Auditor to inspect the same diff from different angles.
  • Flexible Diff Targeting: Supports uncommitted changes, staged changes, branch diffs, commit ranges, and user-provided diffs or file lists.
  • Structured Triage and Follow-Through: Deduplicates findings, classifies them into patch, decision_needed, defer, or dismiss, and can persist review results into story and sprint tracking files.
  • Use Case: Ideal when a developer wants a high-signal review of a feature branch against its story or spec before marking work complete.

Quick Start

Ask the AI to run code review on your current branch or on a specified diff and provide the related story or spec file if you want acceptance auditing.

Frequently Asked Questions about bmad-code-review

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

FAQPage Schema
How do I run an adversarial code review on a git diff?

To run an adversarial code review on a git diff, the Skill constructs a diff from uncommitted changes or commit ranges, applies parallel review roles, deduplicates findings, and triages them into actionable outcomes.

What is the best way to check if a pull request meets its original spec?

Checking if a pull request meets its spec is handled by an acceptance auditor role that inspects the diff against provided story or spec files to validate implementation completeness and flag mismatches.

Can I review staged and uncommitted code changes before creating a pull request?

Yes, you can review staged and uncommitted changes before creating a pull request. The Skill targets flexible diff inputs including staged edits and uncommitted modifications to identify hidden defects early.

How does edge case detection work during a branch diff triage?

Edge case detection during branch diff triage works by running a dedicated Edge Case Hunter role that inspects the diff from an adversarial perspective to find risky scenarios missed by normal reviews.

What are the limitations of automated spec validation for feature branches?

Automated spec validation for feature branches requires loading optional spec and context files to function properly, and it classifies unresolved issues into defer or decision_needed categories rather than auto-fixing them.