sxhxliang/mcp-access-point
MCP Access Point is an open-source, self-hosted proxy server that converts any existing REST API or web service into an MCP (Model Context Protocol) server on the fly. It acts as a middle layer, accepting MCP client connections and translating them into HTTP requests to the upstream API, then formatting the responses back into MCP-compliant outputs. This allows AI agents and large language models to interact with legacy or third-party services without requiring any code modifications to those services. The tool leverages the OpenAPI specification to automatically discover API endpoints, parameters, and data models, generating corresponding MCP tools and resources dynamically. It supports complex configurations, including routing rules, SSL termination, load balancing across multiple upstreams, and global request/response transformation rules. An admin dashboard provides real-time monitoring of connected services, request logs, and configuration status. Built in Rust on top of the high-performance Pingora proxy framework, MCP Access Point is designed for reliability and low latency. It can be deployed as a standalone binary or via Docker, and is suitable for both local development and production environments. Its design emphasizes ease of use: point it at an API and it immediately makes that API accessible to any MCP-compatible client. By decoupling the MCP interface from the underlying service implementation, it enables rapid integration of diverse APIs into agent workflows, supporting use cases from automating internal tools to enabling natural language access to web services.
- Category
- Developer Tools
- Pricing
- Open Source
Problem solved
Developers and organizations often have existing web services that they want to make accessible to AI agents via the Model Context Protocol, but retrofitting MCP support into each service is time-consuming and error-prone. MCP Access Point eliminates this hurdle by providing a proxy that automatically wraps APIs with an MCP interface, requiring no code changes and minimal configuration. It solves the interoperability gap between traditional REST APIs and the emerging MCP ecosystem.
Use cases
- Expose a legacy REST API to an AI agent for automation without modifying the API's code
- Rapidly prototype MCP integrations by pointing at OpenAPI specs
- Provide a single access point for multiple backend services, consolidating them into one MCP server
- Securely expose internal APIs to external AI clients with SSL termination and routing rules
- Monitor and manage MCP services via an admin dashboard for debugging and configuration updates
Audience
- Developers
- ML Engineers
- Hobbyists
- Researchers
Pros
- Instant MCP integration without code changes
- High performance with Rust and Pingora
- Automatic OpenAPI parsing for rapid setup
- Built-in admin dashboard for monitoring
- Supports SSL, routing, and load balancing
Cons
- Requires deploying and managing an additional proxy service
- Dependent on accurate OpenAPI specs for full feature coverage
- Relatively new project with limited community
Official website: https://github.com/sxhxliang/mcp-access-point
Canonical LambdaBase page: https://www.lambdabase.com/tools/sxhxliang-mcp-access-point