Agents Library MCP Server

MCP Server

This is a Python project designed to serve as an MCP (Multi-Cloud Platform) server. It utilizes FastAPI for the web framework and Uvicorn as the ASGI server. The project also includes an agents-library for managing agent-related rules and prompts.

: Technologies Used

• Python
• FastAPI: Web framework for building APIs.
• Uvicorn: ASGI server.
• mcp: Multi-Cloud Platform SDK.

: Project Structure

• Dockerfile: Used for containerizing the application.
• compose.yaml: Used for running the application with Docker Compose.
• requirements.txt: Lists Python dependencies.
• app/:
  • server.py: The main application server.
• agents-library/:
  • dev_rules.agents.md: Development-related agent rules.
  • security_checks.agents.md: Security-related agent checks.
  • common_prompts.agents.md: Common prompts for agents.

: Getting Started

To get started with this project, you need to have Python 3.10+ and Docker installed on your system.

Prerequisites

• Python 3.10+
• Docker
• pip

Installation

1. Create and Activate Virtual Environment:
   python3 -m venv venv source venv/bin/activate
2. Install Dependencies:
   pip install -r requirements.txt

: Building and Running

You can run the server using Docker Compose or directly with Uvicorn.

Using Docker Compose

To run the server with Docker Compose, use the following command:

Using Uvicorn

To run the server with Uvicorn, use the following command:

: Development Conventions

• Virtual Environments: Always use a virtual environment for dependency management.
• Dependencies: All Python dependencies should be listed in requirements.txt.

: API Endpoints

The following API endpoints are available:

• POST /test/call_tool: Test endpoint for direct tool invocation.
• POST /test/read_resource: Test endpoint for direct resource invocation.

The MCP server also exposes the following tools:

• get_agents_instructions: Retrieves a specific AGENTS.md file for providing AI with instructions and context.
• list_agents_instructions: Lists all available AGENTS.md files.

: Adding to gemini-cli

To add this server to gemini-cli, you need to edit your settings.json file. You can find this file in ~/.gemini/settings.json (user settings) or in .gemini/settings.json (project settings).

Add the following to your settings.json file:

Using the mcp tool

Once the mcp-server is configured in gemini-cli, you can use the mcp tool to interact with the server. For example, to list all available agent instructions:

To retrieve a specific agent instruction file:

See reference.

: License

This project is licensed under the Apache License 2.0.

: Author

This project was started in 2025 by Nicholas Wilde.