Hippycampus
@cromwellian
An (eventually) secure open-source MCP Server to turn any REST endpoint into MCP resources automatically
Overview
What is Hippycampus?
Hippycampus is a LangChain-based CLI and MCP server that supports dynamic loading of OpenAPI specifications and integration with Langflow for visual workflow creation. It is intended for developers who want to combine OpenAPI-driven tools with Langchain agents and Langflow’s graphical interface.
How to use Hippycampus?
Install with uv pip install -e . and set the GOOGLE_API_KEY environment variable. Run in CLI mode with uv run hippycampus-cli or as an MCP server with uv run hippycampus-server --transport sse --port 8000. For Langflow integration, start the MCP server first, then run uv run langflow run and configure the provided custom components.
Key features of Hippycampus
- Dynamic loading of OpenAPI specifications into LangChain agents
- CLI mode for debugging without the MCP server
- MCP server with SSE transport on a configurable port
- Custom Langflow components: OpenApi Service and Hippycampus MCP Server
- Supports testing with a fixed XKCD OpenAPI spec
Use cases of Hippycampus
- Load and test arbitrary OpenAPI specs via a LangChain-powered CLI
- Build Langflow workflows that call external APIs through an MCP server
- Debug OpenAPI parsing and agent interactions in a terminal before deploying to Langflow
- Create visual flows that combine multiple OpenAPI services
FAQ from Hippycampus
What are the prerequisites for running Hippycampus?
Python 3.12.9, the UV package manager, a Google AI Studio API key, and optionally Langflow for visual workflows.
How do I start the MCP server?
Run uv run hippycampus-server --transport sse --port 8000 from the project root.
Why do I see authentication errors?
Ensure the GOOGLE_API_KEY environment variable is set correctly.
How do I use the custom Langflow components?
In the Langflow UI, locate the OpenApi Service and Hippycampus MCP Server components. Configure the OpenApi Service with a valid OpenAPI URL and the MCP Server with http://localhost:8000/sse.
What should I do if the custom components are missing in Langflow?
Verify that LANGFLOW_COMPONENTS_PATH points to the correct langflow/components directory, then restart the Langflow server. If they still don’t appear, try restarting again.
