This is a small AI-powered FastAPI project that interacts with the Rick and Morty API, designed to process free-form user text (like "Tell me about an episode with Blips") and extract meaningful queries using an MCP-style endpoint (Model Context Protocol).

📚 Built as part of my personal learning journey into Python, AI agent architecture, and API design.
🧑‍💻 GitHub: unix14

• 🔍 Free-text AI endpoint (/mcp) for extracting intent and arguments from natural language.
• 🎥 Filters episodes by name keywords and returns air dates in a friendly format.
• 👤 (WIP) Character search functionality.
• 📦 Cleanly loads secrets (e.g., API keys) from .env using a separate config file.
• ⚡ Uses FastAPI and httpx for async performance.

You can access a live preview of the Rick and Morty MCP Server running here.

Feel free to test the API and explore its features through this URL.

POST /mcp { "messages": [ { "role": "user", "content": "Give me details about an episode with something like Rixty" } ], "data_requests": [ { "name": "get_episodes_by_name", "arguments": { "name_keyword": "Rixty" } } ] }

{ "S01E08: Rixty Minutes was aired in 17.03.2014" }

• Python 3.11+
• FastAPI
• httpx (async HTTP client)
• python-dotenv
• Uvicorn

rick-and-morty-api-mcp/ ├── main.py # FastAPI app with /mcp endpoint ├── config.py # Loads API keys and env variables ├── .env # Local secret store ├── test_*.py # Simple test scripts └── README.md 

Create a .env file in the root:

MY_API_KEY=your-key-here 

Then in config.py, it's loaded like this:

from dotenv import load_dotenv import os load_dotenv() API_KEY = os.getenv("MY_API_KEY")

pip install -r requirements.txt uvicorn main:app --reload

Test it on: http://localhost:8000/docs

I built this project as part of my self-guided study in:

• Natural Language to structured data pipelines
• AI-oriented API architecture (MCP-style thinking)
• Async APIs and Python backend development

It’s a stepping stone towards building intelligent agents that understand human intent through structured context.

• Integrate OpenAI or LLMs to automatically convert user messages to data_requests
• Add more endpoint support (characters, locations, episodes, etc.)
• Improve fuzzy matching and scoring on episode names
• Add a simple front-end for fun!

If you're curious about this, feel free to reach out via GitHub: github.com/unix14