Blind Auditor MCP server

Blind Auditor is a small helper that sits between your AI and the code it writes. Instead of your AI handing you code right away, this helper makes it stop, review its own work against a rulebook you set up, and only then show you the result. It is handy if you work with an AI coding assistant and keep finding the same mistakes in what it produces.

What this MCP server does

You ask your AI to write some code. Before showing it to you, the AI sends its draft to Blind Auditor. Blind Auditor then tells the AI to switch into an auditor role and check the code line by line against the rules you wrote in a file called rules.json. If the code scores below the threshold, or breaks a critical rule, the AI has to fix it and try again. Only when the audit passes does the code get released to you.

Click to zoom

What you can do with it

• Make your AI review its own code before you see it
• Enforce your own coding rules, like no hardcoded passwords
• Catch critical mistakes before they reach you
• Set a minimum quality score the code has to reach
• Limit how many times the AI can retry before giving up
• Reset the audit session when you want a fresh start

Try asking your AI

• “Write me a Python function that reads a CSV file and returns the average of a column”
• “Add a login form to my web page and show me the code”
• “Fix the bug in this function and give me the corrected version”
• “Write a small script that renames all the files in a folder”

What it gives back to you

You get the final code only after it has passed the audit, so what shows up in your chat is the cleaned-up version. Along the way, the AI may mention the issues it found and fixed. If the audit keeps failing after the maximum number of tries, the code is released anyway with a warning so you know it did not fully pass.

Before you start

Who it's for

People who use an AI coding assistant and want it to follow their team's coding standards and catch its own mistakes before handing over code.