rcode-review

Review code diffs and report severity-ranked findings with a verdict.

2|1|Updated Apr 10, 2026
One-click install
npx skills add https://github.com/hanzlahabib/rcode --skill rcode-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rcode-review
Source: https://github.com/hanzlahabib/rcode/tree/main/rcode/skills/actions/4-implementation/rcode-code-review
Command: npx skills add https://github.com/hanzlahabib/rcode --skill rcode-review

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It turns manual code review into a structured adversarial process that finds real bugs, regressions, and spec drift before merge.

Core Features & Use Cases

  • Parallel review layers: Combines blind, edge-case, and acceptance-focused analysis to catch issues from different angles.
  • Actionable reporting: Produces severity-ranked findings with a clear verdict such as APPROVE, REQUEST CHANGES, or BLOCK.
  • Flexible review sources: Works with staged changes, uncommitted work, branch diffs, commit ranges, and story-driven implementation reviews.

Quick Start

Ask the rcode review skill to review the current diff and report actionable findings.

Frequently Asked Questions about rcode-review

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

FAQPage Schema
How does adversarial code review find bugs before a pull request is merged?

Adversarial code review applies parallel blind, edge-case, and acceptance-focused analysis layers to branch diffs, deduplicating overlaps to catch real bugs, regressions, and spec drift before merge.

Can I review uncommitted work and staged changes without creating a commit?

Yes, you can review uncommitted work and staged changes directly. The review process requires diff input plus optional spec and context documents to triage findings by severity.

How do I triage code review findings by severity for a branch diff?

To triage code review findings by severity, provide the branch diff along with optional spec and context documents. The review process deduplicates overlaps and produces severity-ranked actionable findings.

What is the difference between standard and adversarial code review for spec validation?

Unlike standard review, adversarial code review combines multiple parallel analysis layers to scrutinize implementation stories against spec documents, producing a clear final verdict like APPROVE, REQUEST CHANGES, or BLOCK.

Does this code review process work with commit ranges and story-driven implementation reviews?

Yes, the review process works with commit ranges and story-driven implementation reviews. It requires diff input plus optional spec and context documents to produce a final verdict.

When should I expect a BLOCK verdict from a pull request code review?

A BLOCK verdict occurs when the adversarial review of pull requests or branch diffs surfaces critical findings. It triages severity by comparing diff input against optional spec and context documents.