crapkit MCP server

crapkit is a helper for programmers that looks at every function in a code project and gives it a score. The score combines how complicated the function is with how much of it your tests actually run. It then shows you the worst ones first, so you know what to fix next. It is handy if you write code and want an honest list of trouble spots instead of guessing.

What this MCP server does

You ask your AI something like which function in my project is the riskiest. Your AI uses this helper, which runs crapkit on your project. crapkit reads each function, measures how complex it is, and joins that with the test coverage your own test command already produced. It ranks the functions by risk, then hands the list back to your AI, which explains it to you in plain words.

Click to zoom

What you can do with it

• List the riskiest functions in your project, worst first
• Show which lines inside a function are not covered by tests
• Point out near-duplicate functions that could be merged
• Check whether a change you made adds new risky functions
• Explain why a particular function scores badly
• Compare the current score of a function with its saved baseline

Try asking your AI

• “Which function in this repo is the worst to fix next, and why?”
• “Show me the uncovered lines in the top three risky functions”
• “Did my last change add any new functions over the complexity ceiling?”
• “Are there near-duplicate functions in this file I should merge?”

What it gives back to you

You get back a short ranked list, usually with the function name, its file and line, its complexity number, its coverage percentage, and its score. Your AI often adds a sentence or two explaining what makes each one risky. If you asked about uncovered lines, you get those line numbers too. Nothing in your code is changed by this server; it only reads and reports.

Before you start

Who it's for

Developers and tech leads who want a clear, ranked list of which functions in their codebase deserve attention first.