code-review

Review code changes against team conventions for naming, TypeScript, security, and testing.

Updated Jul 4, 2026
One-click install
npx skills add https://github.com/Assamir/ai-toolkit --skill code-review-assamir
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/Assamir/ai-toolkit/tree/main/packages/ai-toolkit/skills/code-review
Command: npx skills add https://github.com/Assamir/ai-toolkit --skill code-review-assamir

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Code reviews are often inconsistent because reviewers apply personal preferences instead of shared standards. This Skill applies a fixed set of team engineering conventions to every review, producing structured findings grouped by severity with a single clear recommendation. ## Core Features & Use Cases - Convention-Based Review: Evaluates changes against explicit rules for naming, error handling, TypeScript usage, function design, security, and testing. - Severity-Grouped Findings: Reports issues as Critical, Warning, or Suggestion, each with a file:line reference and a concrete fix. - Single Verdict Output: Ends every review with exactly one recommendation: APPROVE, REQUEST CHANGES, or NEEDS DISCUSSION. - Use Case: Before merging a pull request, ask for a review of the diff to catch missing error handling, any types, unparameterized SQL, or weak test assertions against your team's standards. ## Quick Start Review my staged changes against the team engineering conventions and give me a severity-grouped report with a final recommendation.

Frequently Asked Questions about code-review

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

FAQPage Schema
How do I review a pull request against team coding standards?

Provide the diff, PR, staged files, or a named path and ask for a code review. The review evaluates each changed file against conventions for naming, error handling, TypeScript, security, and testing, then reports findings grouped by severity.

What does an automated code review check in TypeScript projects?

It checks for unjustified any types, interface versus type usage, unknown with type guards for external data, discriminated unions for state modeling, and descriptive generic parameter names, alongside naming, error handling, and security rules.

How are code review findings organized in the output?

Findings are grouped by severity: Critical for must-fix issues like security or data loss, Warning for convention violations, and Suggestion for optional improvements. Each includes a file:line reference and a concrete fix.

Can this review catch security issues like SQL injection?

Yes, the security category flags hardcoded secrets, missing input validation at system boundaries, non-parameterized SQL statements, and API responses that leak stack traces or internal paths.

When should a code review end with NEEDS DISCUSSION instead of APPROVE?

Use NEEDS DISCUSSION when findings involve trade-offs or ambiguous intent that cannot be resolved by conventions alone. APPROVE fits clean changes, while REQUEST CHANGES applies when Critical or significant Warning findings exist.