FastMCP context servers
- Python
- MCP
- AI
An agent that has to infer a project’s structure, conventions and history from scratch every session wastes tokens re-discovering things a human teammate would already know. These are small, focused MCP servers built with FastMCP that hand that context over directly: repository layout, documentation, coding conventions, and task-tracking (one of them backs the very TODO list used to plan this site’s content).
Each server does one job well rather than trying to be a do-everything assistant: read the project’s actual state, describe it in a form an agent can act on immediately, and stay out of the way otherwise. It’s unglamorous infrastructure, but it’s the difference between an agent that asks the right question once and one that asks it every session.
This is private, internal tooling rather than a public repository — it’s built to run alongside whichever client project it’s supporting, not to be a standalone product.