cc-sensei MCP server

cc-sensei is a small helper that gives your AI a map of how Claude Code works inside. It comes with written notes about 32 parts of Claude Code, plus the actual source code, all stored on your own computer. It is handy if you are building your own AI agent and want to see how a real one was put together.

What this MCP server does

You ask your AI a question in plain words, like how Claude Code handles long conversations. Your AI passes the question to cc-sensei. cc-sensei looks through its 32 module write-ups and finds the matching ones. It sends back a short explanation, the design choices behind it, and the file names where the code lives. If you want, it can also open the actual source file and show you the lines.

Click to zoom

What you can do with it

• List all 32 modules that make up Claude Code
• Search the notes in plain words and get a brief, standard, or deep answer
• Drill into one module to see its responsibilities, decisions, and principles
• Trace one topic, like prompt cache, across all modules
• Pull out reusable design patterns you can copy
• Read the actual Claude Code source files with line numbers

Try asking your AI

• “Use cc-sensei to tell me how Claude Code compacts long conversations.”
• “Which modules in Claude Code deal with prompt cache?”
• “Show me the design principles behind the context engineering module.”
• “Open the file services/compact/microCompact.ts and show me the first 100 lines.”

What it gives back to you

You get answers in your chat window: short explanations, lists of modules, comparison tables, and design principles with the file names where each idea lives. When you ask for source code, it shows the file with line numbers so you can read it yourself. Everything comes back as text, and it stays short on purpose so your chat does not get flooded.

Before you start

Who it's for

People building their own AI agents who want to learn from how Claude Code was designed, without reading hundreds of thousands of lines of source themselves.