Entroly MCP server

Entroly is a helper that sits between your code and your AI assistant. Instead of handing the AI your whole project, it picks out just the parts that matter for your question. It is handy if you work with a large codebase and your AI keeps getting slow, expensive, or confused.

What this MCP server does

You ask your AI a question about your code. The AI asks Entroly for help. Entroly looks through your project on your own machine and picks the fragments that fit your question and a token budget. It hands those pieces back to the AI, along with a receipt that says what was kept and what was left out. You get an answer that is based on the selected parts, and you can check the receipt to see the reasoning.

Click to zoom

What you can do with it

• Select the most relevant parts of a large codebase for a specific question
• Keep a local receipt of what was included and what was omitted
• Recover the exact original text behind any selected fragment
• Run a local check for hallucinations in the AI's answer
• Route simpler questions to a cheaper model when appropriate
• Reduce the number of tokens sent to the AI to lower costs
• Work with editors like VS Code, Cursor, and Claude Code

Try asking your AI

• “Find the code that handles user login and explain how it works.”
• “Which files are involved in processing payments?”
• “Summarize the rate limiting logic in this project.”
• “Show me the parts of the codebase related to authentication and authorization.”

What it gives back to you

You get a focused answer from your AI, built from the selected parts of your code. Entroly also gives you a receipt that lists what was kept, what was left out, and why. If you want, you can ask for the exact original text behind any fragment. The receipt stays on your machine so you can review it later.

Before you start

Who it's for

Developers and technical teams who work with large codebases and want their AI assistant to be faster, cheaper, and more accurate.