MCP Server
by eagurin
MCP Server is a FastAPI-based implementation of the Model Context Protocol (MCP), providing a standardized interface for interaction between LLM models and applications. It offers features like high-performance API, complete MCP support, monitoring, and extensibility.
Last updated: N/A
What is MCP Server?
MCP Server is a Model Context Protocol (MCP) implementation built on FastAPI that facilitates standardized communication between Large Language Models (LLMs) and applications by providing resources, tools, prompts and more via an API.
How to use MCP Server?
To use MCP Server, clone the repository, install Poetry (if you don't already have it), install dependencies using Poetry, and then run the server using the provided command. Access the API documentation via Swagger UI, ReDoc, or GraphQL Playground.
Key features of MCP Server
High-performance API based on FastAPI
Full MCP support
Monitoring and metrics via Prometheus and Grafana
Extensibility for adding new tools
GraphQL API for flexible data access
WebSocket support for real-time interaction
Semantic Search via Elasticsearch integration
Caching via Redis
Use cases of MCP Server
Integration of LLMs into applications.
Providing a standardized interface for LLM interaction.
Creating and managing tools for LLMs.
Monitoring and analyzing LLM usage.
FAQ from MCP Server
What is Model Context Protocol (MCP)?
What is Model Context Protocol (MCP)?
MCP is a standard for providing context and tools to LLMs, enabling them to interact with external systems and data.
How do I add a new tool to MCP Server?
How do I add a new tool to MCP Server?
MCP Server is designed to be extensible, add your tool to the tools/. See example_tool.py as reference.
How do I monitor the performance of MCP Server?
How do I monitor the performance of MCP Server?
MCP Server exposes metrics in Prometheus format at the /metrics endpoint. You can use Prometheus and Grafana to monitor the server's performance.
How do I use the GraphQL API?
How do I use the GraphQL API?
The GraphQL API is available at /graphql. You can use the GraphQL Playground to explore the schema and execute queries.
Does it support docker-compose?
Does it support docker-compose?
Yes. The documentation provides an example docker-compose to build and run the service locally.