code-review

Reviews code changes in diffs, commits, and branches with evidence-based findings and suggested fixes.

Updated Apr 2, 2026
One-click install
npx skills add https://github.com/nico0695/ai-tools --skill code-review-nico0695
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/nico0695/ai-tools/tree/main/skills/generic-skills/code-review
Command: npx skills add https://github.com/nico0695/ai-tools --skill code-review-nico0695

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Reviewing code changes manually is time-consuming and error-prone, especially when assessing cross-module impact, convention compliance, and hidden side effects. This Skill provides a structured, multi-phase review workflow that delivers direct, evidence-based assessments with actionable fixes for every issue found. ## Core Features & Use Cases - Scoped Review Modes: Choose between a light review (fast diff scan for critical issues only) or a deep review (full analysis with cross-module impact, conventions check, and a second-pass side-effect scan). - Parallel Subagent Analysis: Deep mode delegates diff analysis, cross-module impact tracing, and convention checking to independent subagents, then runs a final side-effect scanner to catch issues visible only in combination. - Structured Reports with Fixes: Every finding includes exact file:line location, why it matters, and a suggested fix, with optional before/after code snippets grouped by file. - Use Case: Before merging a feature branch, ask for a deep review of the branch diff to catch security issues, broken logic, convention violations, and cascading side effects across modules. ## Quick Start Review the changes on my current branch compared to main with a deep review and show me suggested fixes for any issues found.

Frequently Asked Questions about code-review

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

FAQPage Schema
How do I review code changes in a git branch before merging?▼

Specify the target branch or commit range, state the intent of the changes, and choose a review depth. The skill reads the diff via git diff or git log, analyzes it for critical issues, and returns a structured report with file:line references and suggested fixes.

What is the difference between a light and deep code review?▼

Light mode is a fast diff scan that flags only critical issues like security flaws, crashes, and broken logic. Deep mode adds cross-module impact analysis, convention checking against repo docs, and a second-pass scan for side effects that emerge from combined changes.

Can this review detect side effects across multiple files?▼

Yes. In deep mode, a dedicated side-effect scanner subagent receives the diff plus all prior findings and looks specifically for cascading state mutations, race conditions, and issues that only appear when changes are combined across files.

Does the code review work without knowing the intent of the changes?▼

No. The skill requires the review target, the intent behind the changes, and a depth level before proceeding. A diff without stated intent is treated as syntax only, since correctness cannot be evaluated without knowing the goal.

What are the limitations of automated diff-based code review?▼

The review is based only on what is visible in the diff and referenced files, so it cannot verify runtime behavior or unexamined code paths. It does not read the entire codebase and explicitly states when something cannot be verified.