explore-python

name	explore-python
description	Research Python/FastAPI MCP server project code - tools, structure, patterns, FastAPI organization, asyncio, _meta, human-in-the-loop, OTel, Auth
context	fork
agent	Explore
allowed-tools	Read Grep Glob Bash

Explore MCP Server (Python)

Explore this project's Python/FastAPI MCP server code.

Orientation — read these first

Before exploring code, check project state:

• CLAUDE.md — conventions, architecture decisions, setup notes
• CHANGELOG.md — what's been implemented, recent changes, what's in progress

Both may be empty early in the project. Note that and proceed to code.

Service Context

• Python >= 3.13, FastAPI >= 0.115
• MCP protocol via FastMCP (fastmcp); tool registration with @mcp.tool() async defs
• _meta field used for human-in-the-loop elicitation / approval flows
• Package manager uv (Astral); build backend hatchling, lockfile: uv.lock
• Linter ruff (target py313); type-check mypy >= 1.14
• External MCP servers in use: context7, GitHub MCP, Serena MCP, rtl (rust token killer)

What to look for

• Tool definitions (@mcp.tool()) — inputs, outputs, _meta handling
• Human-in-the-loop patterns — elicitation requests, approval flows, _meta passthrough
• FastAPI route organization — async handlers, dependency injection, middleware
• Auth patterns — API keys, OAuth, token validation
• Observability — OTel spans, logging, structured output
• Error handling — MCP error codes, HTTP status mapping
• Project structure — how modules are split, where config lives

Task

$ARGUMENTS