review-code

Reviews a diff against standards and spec axes using parallel sub-agents.

Updated Feb 10, 2026
One-click install
npx skills add https://github.com/vesviet/agent-skills --skill review-code-vesviet
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: review-code
Source: https://github.com/vesviet/agent-skills/tree/main/core/skills/repo-ops/review-code
Command: npx skills add https://github.com/vesviet/agent-skills --skill review-code-vesviet

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Code reviews often miss subtle AI-generated defects like phantom validations, silent error swallowing, and mock collusion, and reviewers frequently conflate standards compliance with spec fidelity, letting one axis mask failures in the other. ## Core Features & Use Cases - Two-Axis Independent Review: Evaluates Standards (repo standards plus Fowler smell baseline) and Spec (faithful implementation of the originating issue) as parallel sub-agents, reported side by side without cross-ranking. - Adversarial Anti-Slop Audit: Detects AI bug signatures including hallucinated APIs, async race conditions, context-window truncation, tautological test assertions, and resource leaks in error paths. - Safety Verification: Enforces zero-downtime expand/contract database migrations, backward-compatible API changes, and supply chain checks for typosquatting and unpinned dependencies. - Use Case: Before merging a pull request, run the skill against a fixed point like main to receive separate Standards and Spec findings with exact file paths and line numbers, plus a structured code-review-finding.json for CI gating. ## Quick Start Use the review-code skill to review all changes since main on both the standards and spec axes and report findings side by side.

Frequently Asked Questions about review-code

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

FAQPage Schema
How do I review a git diff against coding standards and a spec?▼

Pin a fixed point such as a commit SHA or branch, capture the diff with git diff <fixed-point>...HEAD, then run the Standards and Spec axes as parallel sub-agents. Each axis reports findings independently under separate headings without cross-ranking.

What is a two-axis code review?▼

A two-axis review evaluates Standards (documented repo standards plus the Fowler smell baseline) and Spec (faithful implementation of the originating issue) as separate, unmerged streams. This separation prevents strong findings on one axis from masking failures on the other.

How do I detect AI-generated bugs in code reviews?▼

Apply adversarial heuristics targeting AI defect signatures: hallucinated APIs, missing awaits, race conditions, dropped switch branches, over-defensive null coalescing, and tautological test assertions. Verify function signatures against pinned dependency versions and trace all error paths.

Can I run this review if there is no spec or issue?▼

Yes. If no spec is found after checking commit references, user-provided paths, and docs directories, the Spec axis still runs and reports "no spec available". The Standards axis proceeds normally with repo standards and the smell baseline.

What database migration changes fail this review?▼

Single-phase destructive changes like dropping or renaming a column in one deploy are rejected. Migrations must follow the zero-downtime Expand/Contract pattern: add the new column nullable, dual-write and backfill, then contract in a later release.

When should I not use this review skill?▼

Skip it for informal exploratory walkthroughs that do not gate a merge, since structured contract emission is unnecessary there. It also duplicates work already enforced by automated linters and type checkers, which the review explicitly skips.