FastAPI MCP Production Kit
A local-first FastAPI and MCP safety kit for turning internal HTTP APIs into controlled MCP tools, with per-tool scopes, quotas, audit events, and default-deny web-access boundaries.
README
FastAPI MCP Production Kit
A local-first FastAPI and MCP safety kit for turning internal HTTP APIs into controlled MCP tools.
Most MCP examples show how to expose a tool. This repo focuses on what teams need before agents can use those tools safely: local credentials, per-tool scopes, quota checks, audit events, web-access boundaries, fallback decisions, and a quickstart that works without paid services.
This kit helps you answer:
- Which FastAPI capabilities should become MCP tools?
- How do tool calls prove identity before doing work?
- How do different tools get different scopes?
- How do quotas and audit events work before remote deployment?
- How do web-access tools default to deny instead of arbitrary outbound access?
Ships today:
- FastAPI app factory with
/healthz, tool discovery, demo token, and tool-call endpoints - Local MCP-style tool dispatcher with three tools
- HMAC-signed local demo tokens
- Per-tool scope checks
- Deterministic in-memory quotas
- Structured audit events and JSONL fixture generation
- Default-deny web-access fixture boundary
- Provider fallback decision record helper
- Pytest coverage for auth, scopes, quotas, audit, boundaries, fallback, and tool calls
- Public boundary scan script
- Production docs map for security, scopes, audit, quotas, deployment, web access, fallback, observability, and troubleshooting
Quickstart
python3 -m venv .venv
source .venv/bin/activate
python -m pip install -e '.[dev]'
pytest
python examples/local-only-demo/demo_client.py
Run the FastAPI app:
uvicorn prodkit_mcp.app:app --reload
List available tools:
curl -s http://127.0.0.1:8000/tools
Create a demo token:
curl -s http://127.0.0.1:8000/demo/token \
-H 'content-type: application/json' \
-d '{"subject_id":"local-developer","scopes":["project:read","docs:search","web:fetch"]}'
Call a tool:
curl -s http://127.0.0.1:8000/tools/read_project_status \
-H "authorization: Bearer $ACCESS_TOKEN" \
-H 'content-type: application/json' \
-d '{"arguments":{}}'
Generate audit fixtures:
python scripts/generate_audit_fixtures.py
Run the public boundary scan:
python scripts/scan_public_boundary.py
Tool Model
| Tool | Scope | Purpose |
|---|---|---|
read_project_status |
project:read |
Reads a synthetic project status record |
search_docs_fixture |
docs:search |
Searches bundled documentation fixtures |
fetch_allowed_page |
web:fetch |
Fetches only reviewed safe page fixtures |
What This Is Not
This is not a hosted MCP platform, a broad server directory, a production security review, or an arbitrary web-access tool. It is a local reference kit for making MCP tool exposure easier to reason about before remote deployment.
Production Guides
- Production docs map
- Security model
- Scopes
- Audit events
- Quotas
- Deployment
- Web-access boundaries
- Provider fallback
- Observability
- Troubleshooting
Optional MCP SDK Adapter
The default quickstart uses the local dispatcher so the safety path is easy to
test in CI. If you install the optional mcp extra, prodkit_mcp.mcp_server
can create a FastMCP server wrapper around the same tools.
python -m pip install -e '.[mcp,dev]'
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.