CodeSafer MCP server

CodeSafer is a small helper that checks code for hidden problems, the kind that look fine on screen but do something else when they run. It is handy if you use AI to write or review code and want a second pair of eyes before you trust it. You add it to your AI app once, then just ask your AI to scan things.

What this MCP server does

You ask your AI something like scan this file for hidden security issues. Your AI passes that request to CodeSafer, which runs on your own computer. CodeSafer looks through the file, or a whole folder, using a set of built-in rules and a trained model. It then hands back a list of findings with severity, line numbers, and advice on how to fix them. Nothing is uploaded anywhere; the checking happens locally.

Click to zoom

What you can do with it

• Scan a single file for hidden or malicious code patterns
• Scan a whole folder of source files at once
• Check AI rules files like .cursorrules or CLAUDE.md for planted instructions
• Check package.json for typosquatted or suspicious dependencies
• Run a deeper AI analysis on a piece of code with a confidence score
• Ask for a plain explanation of a threat category and how to fix it

Try asking your AI

• “Scan this file for hidden security issues.”
• “Check the dependencies in package.json for typosquatting.”
• “Scan .cursorrules for a rules-file backdoor.”
• “Run a deep AI analysis of src/auth.ts.”

What it gives back to you

You get a list of findings in the chat, each with a severity level, the line number where it was found, and a short note on what is wrong. For deeper checks, you also get a confidence score from the AI analysis. If nothing suspicious turns up, it simply tells you the file looks clean. You can then ask follow-up questions like what a Trojan Source attack is and how to fix it.

Before you start

Who it's for

Developers and anyone who uses AI to write or review code and wants a quick safety check before running it.