My Own Assistant MCP Server

What is My Own Assistant?

My Own Assistant is an experimental project to simplify creating a fully integrated AI assistant using the Model Context Protocol (MCP). It is designed to be fully self‑hosted by bringing your own OpenAI compatible API, taking advantage of models like Qwen3‑30B‑A3B. The repository is built on a FastAPI template that supports CPU/GPU acceleration and includes common development tooling.

How to use My Own Assistant?

Clone the repository and replace all instances of py‑api‑template with your own project name. Optionally rename the src folder. Ensure the virtual environment is active and poetry install has been run. Use poe dev to start the development server, poe build {version} to build a Docker image, and poe prod to run the image locally. VSCode’s debug task (F5) is available for debugging.

Key features of My Own Assistant

• Fully self‑hosted with your own OpenAI compatible API
• Built on a FastAPI template with automatic CPU/GPU detection
• Package management via Poetry, formatting and linting via Ruff
• Pre‑configured VSCode debugging tasks
• Support for Dev Containers clone‑in‑volume feature
• Docker image build and deployment workflows

Use cases of My Own Assistant

• Creating a personal, self‑hosted AI assistant integrated with MCPs
• Experimenting with high‑efficiency models such as Qwen3‑30B‑A3B
• Using an OpenAI compatible API to power a custom assistant
• Starting a FastAPI‑based MCP server project with ready‑to‑use tooling

FAQ from My Own Assistant

—