Analysis updated 2026-05-18
Review a code snippet or file for bugs and security issues before merging
Get a blunt, direct review of a git diff or commit instead of a softened one
Catch SQL injection and other OWASP-style vulnerabilities in your code
Choose a harshness level from gentle to maximum bluntness for reviews
| frangelbarrera/code-review-agent-mcp | 1ncendium/aibuster | aaronmayeux/ha-hurricane-tracker | |
|---|---|---|---|
| Stars | 5 | 5 | 5 |
| Language | Python | Python | Python |
| Setup difficulty | easy | moderate | easy |
| Complexity | 2/5 | 3/5 | 2/5 |
| Audience | developer | ops devops | general |
Figures from each repo's GitHub metadata at analysis time.
Requires Python 3.10+ and an MCP-compatible client such as Claude Desktop or Cursor.
Code Review Agent MCP is a tool that plugs into AI coding assistants like Claude Desktop or Cursor and gives them the ability to review code the way a strict, experienced maintainer would, without softening the message. It connects using MCP, a protocol that lets AI assistants call outside tools, so once installed you can ask your assistant to review a code snippet, a whole file, a git diff, or a specific commit, and it will return findings directly through the conversation. Each finding gets a severity label of critical, major, minor, nit, or clean, points to the exact line number involved, explains what is wrong and how to fix it, and ends with a clear verdict of either ship it or do not merge until the problems are fixed. The project was built specifically to avoid the overly polite tone many AI tools default to, where criticism gets buried under phrases like great work but or you might want to consider. Instead it aims to sound like reviewers on serious open source projects such as the Linux kernel, who state problems plainly because their goal is a better result, not a nicer conversation. To keep that blunt tone consistent, the project uses three layers working together: a detailed system prompt with rules against hedging and false praise, a post processing step that strips out dozens of softening phrases from the output, and a validator that checks each review actually includes severity labels, line citations, and a verdict. There are also four harshness levels to choose from, ranging from gentle to a maximum bluntness kernel maintainer mode. The tool specifically watches for security issues like SQL injection alongside general bugs such as logic errors, race conditions, and resource leaks, and it is designed to explicitly say when code is clean rather than inventing problems that are not there. It includes sandboxing so file access stays limited to the current project folder and refuses to touch sensitive paths. It is installed through pip, written in Python, and released under the MIT license.
An MCP server that plugs into AI coding assistants to give blunt, line-specific code reviews with clear severity labels and a ship or do not merge verdict.
Mainly Python. The stack also includes Python, MCP, pip.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.