code-review

Performs multi-pass static analysis detecting defects, tainted data flows, and concurrency issues across language stacks.

1|Updated Jun 27, 2026
One-click install
npx skills add https://github.com/vickysrawat/AI-Assisted-development --skill code-review-vickysrawat
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: code-review
Source: https://github.com/vickysrawat/AI-Assisted-development/tree/main/skills/code-review
Command: npx skills add https://github.com/vickysrawat/AI-Assisted-development --skill code-review-vickysrawat

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Manual code review misses deep inter-procedural defects like tainted data flows, null dereferences, resource leaks, and race conditions that span function boundaries. This Skill runs a deterministic analyzer layer plus a three-pass LLM scan architecture to find real defects with concrete, copy-pasteable fixes. ## Core Features & Use Cases - Three-Pass Scan Architecture: Combines deterministic analyzers (C# build warnings, eslint, web.config checks), structured rule-based checkers, specialized persona reviews (reliability, concurrency, API contracts), and a free-flow adversarial pass for architectural risks. - Stack-Agnostic Detection: Detects .NET, TypeScript/Angular, Java/Spring, Python, Go, and VSTO codebases and loads only matching checker files, including per-project .NET Framework vs modern .NET lensing. - Ledger and Baseline Tracking: Writes findings to a persistent ledger with fingerprints, baselines, and capability-aware reconciliation so only new or touched findings gate commits. - Use Case: Run a full review before a release to get CID-numbered findings with event paths, vulnerable code snippets, corrected code, and CWE/OWASP references, plus an HTML report. ## Quick Start Ask the assistant to run a code review on the changed files in this repository and list all critical and high severity findings with fixes.

Frequently Asked Questions about code-review

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

FAQPage Schema
How do I run a static code analysis review on my repository?

Invoke the code review skill with a scope flag such as --changed for modified files, --pr for branch diffs, or --full for the entire codebase. It detects your language stack automatically and loads only the matching checker files.

What languages does this static analysis support?

It supports C#/.NET (Framework through modern .NET), TypeScript/Angular/Node.js, Java/Spring Boot, Python (FastAPI, Django, Flask), Go, and VSTO Office add-ins. Polyglot repos load all matching checker files, and mixed .NET solutions get per-project version-aware analysis.

How does the review avoid re-reporting old findings?

Findings are fingerprinted and stored in a ledger with a non-gating baseline from the first deterministic run. Later scans only surface findings new versus baseline or in files you touched, and each pass dedupes against prior passes.

Can I use this in a CI pipeline without interactive prompts?

Yes, pass --ci or have another gate invoke it with --changed, which skips the interactive menu and runs a headless cache-aware scan. The checkin commit gate uses the lighter changed-files path automatically.

What output does the code review produce?

It generates a Markdown report and a self-contained HTML report in a CodeReviews folder, plus a persistent ledger. Every finding includes a CID, checker name, event path, vulnerable code snippet, corrected fix snippet, and CWE/OWASP references.