mcp-all s

This MCP setup acts as a central reference server collection that demonstrates how MCP can orchestrate access to a wide range of tools and data sources. The mcp-all server is designed to be run as a Docker container that encapsulates multiple reference implementations (as listed in the repository) and exposes a unified interface for LLMs to query tools, data stores, and services. Use this server to explore capabilities such as document retrieval, web scraping, file system access, database queries, and more, all through a consistent MCP protocol layer. Once running, you can send prompts to the MCP endpoint and request tool invocations, with the server delegating to the appropriate reference implementation under the hood.

To interact with the server, send standard MCP requests (e.g., tool invocations, tool parameter queries, and result callbacks) to the exposed MCP API endpoint. Depending on the image, endpoints may be accessible on a mapped port (default 8080 or as configured via MCP_ALL_PORT). The server will route requests to the appropriate reference implementation (e.g., AWS KB retrieval, Google Drive access, PostgreSQL read-only queries, and more) based on the prompt and the MCP tool descriptors defined in the container.

Prerequisites:

• Docker installed and running on your machine
• Sufficient disk space to pull and run the container

Step-by-step installation:

1. Pull the MCP all-in-one image (if not already available locally): docker pull aswinthulasir/mcp-all-servers

2. Run the container in interactive mode, exposing the MCP endpoint (adjust port if needed): docker run -it --rm -p 8080:8080 aswinthulasir/mcp-all-servers

3. Verify the server is up by checking logs or hitting the health endpoint (adjust URL based on container docs): curl http://localhost:8080/health

4. (Optional) Set environment variables to customize behavior, such as log level or port, when starting the container: docker run -it --rm -p 8080:8080 -e MCP_ALL_LOG_LEVEL=debug -e MCP_ALL_PORT=8080 aswinthulasir/mcp-all-servers

Notes and tips:

• This repository hosts multiple reference MCP servers; the mcp-all image aggregates these implementations for convenient experimentation.
• If you run into port conflicts, adjust the MCP_ALL_PORT environment variable or the -p flag when starting the container.
• Some reference servers may require additional credentials or API keys (e.g., AWS, Google Drive). Provide these via environment variables or mounted secrets as appropriate for the image you are using.
• Check the container logs for which MCP tools are currently exposed and how requests are routed to individual reference implementations.
• Ensure your LLM or client adheres to the MCP protocol for authenticating and invoking tools; the server acts as a gateway to the underlying reference implementations.