Agents Library MCP Server

# :robot: MCP Server :computer: [](https://github.com/nicholaswilde/mcp-server/actions/workflows/test.yml) [](https://taskfile.dev/#/) An MCP (Multi-Cloud Platform) server that provides a library of reusable agent instructions and scripts to a generative AI model. >[!WARNING] >This project is in a development stage. Features and configurations are subject to change. ## :book: Documentation For comprehensive documentation, please visit the [MkDocs site](https://nicholaswilde.io/mcp-server/). ## :mag: TL;DR The MCP Server uses FastAPI to expose a set of tools that can be consumed by a compatible AI model (like Google's Gemini). It provides a library of standardized instructions (`AGENTS.md` files) and utility scripts (`.sh` files) to enable the AI to perform complex, context-aware tasks. To bootstrap the project ```bash task bootstrap ``` To run the server locally: ```bash task run ``` Add to `gemini-cli` settings: ```json { "mcpServers": { "sharedAgents": { "httpUrl": "http://<ip-address>:8080" } } } ``` ## :mag: Overview This server uses FastAPI to expose a set of tools that can be consumed by a compatible AI model (like Google's Gemini). The primary purpose is to provide the AI with a library of standardized instructions (`AGENTS.md` files) and utility scripts (`.sh` files). This allows the AI to perform complex, context-aware tasks consistently by drawing from a central, version-controlled library. The core components are: - **`app/server.py`**: The FastAPI application that serves the tools. - **`agents-library/`**: The central repository for agent instructions and scripts. ## :balance_scale: License This project is licensed under the [Apache License 2.0](./LICENSE). ## :pencil: Author This project was started in 2025 by [Nicholas Wilde](https://github.com/nicholaswilde/).