ADK MCP Agent
A streaming chat agent using Google ADK and the Model Context Protocol (MCP) Google Maps toolset.
Local Development
-
Install dependencies:
pip install -r requirements.txt -
Set up
.envinapp/with your API keys. -
Run locally:
export SSL_CERT_FILE=$(python -m certifi) uvicorn app.main:app --reload
Project Structure
adk-mcp-agent/
├── app/
│ ├── main.py
│ ├── static/
│ │ └── index.html
│ └── .env
├── agent/
│ └── agents.py
├── requirements.txt
├── Dockerfile
├── README.md
└── .gitignore
- The agent logic is in the
agent/directory (agent/agents.py). - The FastAPI app and UI are in the
app/directory.
Deployment
To test and run the app locally with the UI:
-
Install dependencies:
pip install -r requirements.txt -
Set up environment variables:
- Create a
.envfile inside theapp/directory with your API keys:GOOGLE_GENAI_USE_VERTEXAI=FALSE GOOGLE_API_KEY=your_genai_api_key GOOGLE_MAPS_API_KEY=your_google_maps_api_key
- Create a
-
Run the FastAPI app:
export SSL_CERT_FILE=$(python -m certifi) uvicorn app.main:app --reload -
Access the UI:
- Open your browser and go to http://localhost:8000
- You can now interact with the chat interface and test the agent locally.
License
MIT