API Doctor MCP
Validates OpenAPI documents, JSON Schemas, and JSON payloads. Also compares OpenAPI specs and displays breaking changes.
README
API Doctor MCP
A deterministic, stateless MCP server for validating OpenAPI documents, JSON Schemas, and JSON payloads.
It uses Python libraries only. It does not call an LLM, persist user data, or require login.
Current tools
validate_openapilist_openapi_endpointscompare_openapi_specsview_openapi_breaking_changesview_openapi_validation_cardvalidate_json_schemavalidate_payload
Architecture
MCP transport
-> ServerFactory
-> ToolProvider discovery
-> Handlers
-> Services
-> Validators / Comparators
Important properties:
- Stateless: no database, sessions, cache, or persistent files.
- Open/Closed: add a new
IToolProviderimplementation; discovery registers it automatically. - Dependency Inversion: handlers depend on service interfaces, not concrete libraries.
- Strategy: parsers, validators, and comparison rules are replaceable strategies.
- Factory: application and server creation are centralized.
- Adapter: third-party validation libraries are hidden behind internal interfaces.
- Class-based: application behavior is represented by classes and objects.
- CamelCase Python filenames: used as explicitly requested.
Run locally
uv sync --extra dev
uv run uvicorn app.Main:app --host 0.0.0.0 --port 8085
MCP endpoint:
http://localhost:8085/mcp
Health endpoint:
http://localhost:8085/health
The server uses the MCP Python SDK v2 beta and exposes one Streamable HTTP application. Modern 2026-07-28 requests are sessionless. stateless_http=True also prevents legacy clients from creating retained HTTP sessions.
MCP HTTP usage
The callable MCP endpoint is:
http://localhost:8085/mcp
This server uses the modern stateless MCP HTTP format for protocol version 2026-07-28.
Each MCP request is a single HTTP POST with:
- header
Content-Type: application/json - header
Accept: application/json, text/event-stream - header
mcp-protocol-version: 2026-07-28 - header
mcp-method: <json-rpc-method> - header
mcp-name: <tool-name>fortools/callrequests - JSON-RPC body containing
params._meta
See MCP_API.md for complete request and response examples for:
tools/listvalidate_json_schemavalidate_payloadvalidate_openapilist_openapi_endpointscompare_openapi_specsview_openapi_breaking_changes
MCP Apps
This repo now includes two MCP Apps:
view_openapi_breaking_changesview_openapi_validation_card
It follows the MCP Apps pattern described in the official docs:
- the tool advertises
_meta.ui.resourceUri - the server exposes a
ui://...resource - the host fetches and renders the bundled HTML app in a sandboxed iframe
The bundled app source lives in:
mcp_apps/breaking-change-viewer/
Build the app bundle:
make app-build
The built artifact served by the Python MCP server is:
mcp_apps/breaking-change-viewer/dist/mcp-app.html
Test with basic-host
The easiest local host-level test path is the MCP Apps basic-host.
- Start this server:
make run
- In another folder, clone the MCP Apps repository and start the host:
git clone https://github.com/modelcontextprotocol/ext-apps.git
cd ext-apps/examples/basic-host
npm install
SERVERS='["http://localhost:8085/mcp"]' npm start
- Open:
http://localhost:8080
- Select and call either
view_openapi_breaking_changesorview_openapi_validation_card
When the host calls the tool, it should fetch the ui://... resource and render the inline app. The validation card is intentionally small and is useful as a lightweight MCP Apps demo.
Test
uv run pytest
uv run ruff check .
uv run mypy app
Docker
docker compose up --build
Add a tool without modifying existing code
- Add a service interface and implementation, when needed.
- Add a handler class.
- Add a concrete class implementing
IToolProviderunderapp/ToolProviders. - Give the provider a parameterless constructor.
- The provider-discovery mechanism imports and registers it automatically.
Providers receive the application ServiceContainer during registration.
Production note
The pinned MCP SDK is a prerelease. Upgrade the exact pin deliberately after testing API changes.
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
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.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.