adk-a2a-travel-assistant

# Introduction <img src="assets/ADK_and_A2A.png" width=600></img> This project modifies a previous project: [Local Explorer Assistant](https://github.com/gabrielpreda/local-explorer-assistant) by the followings: - Introduce A2A (Agent to Agent protocol from Google); - Separate Discovery and Routing Agents as separate services; - Exposes Discovery and Routing Agents using A2A; - Keep Composer Agent as SubAgent of the Root Agent, and modify this one to be as well exposed as A2AServer; - Add a A2AClient component; - Small facelift of the Streamlit UI. - Add a logging integrator component, to be able to monitor in one place all the logs from the A2A Servers; this uses the functionality from the new module `logging_ring.py` and a custom Streamlit UI. # Ho to run The instructions are for running the project on local ## Configure your local environment Start by editing the .env file. Copy your `.env.example` as `.env` and replace the placeholders with your actual values. ```bash GOOGLE_GENAI_USE_VERTEXAI=true GOOGLE_CLOUD_PROJECT=<YOUR_PROJECT_ID> GOOGLE_CLOUD_LOCATION=<YOUR_REGION> ``` ## Start the A2A servers Start the Discovery Agent A2A Server (1st Remote Agent): ```bash uvicorn discovery:application --host 0.0.0.0 --port 10020 ``` Start the Routing Agent A2A Server (2nd Remote Agent): ```bash uvicorn agent:application --host 0.0.0.0 --port 10022 ``` Start the Root Agent A2A Server (Host Agent): ```bash uvicorn agent:application --host 0.0.0.0 --port 10022 ``` ## Start the A2A Client The A2A Client exposes a FastAPI service for the UI. It instantiate a A2A Client to connect to the host agent. Start the A2A Client with the following command: ```bash uvicorn main:app --host 0.0.0.0 --port 8000 ``` ## Start the UI (Streamlit) Start the user interface: ```bash streamlit run streamlit_ui.py --server.port 8501 ``` Here is an image of the main client application user interface: <img src="assets/app_ui.png" width=600></img> ## Start the UI for monitoring logs Start the user interface: ```bash streamlit run streamlit__logs_ui.py --server.port 8502 ``` Here is an image of the A2A Agents monitoring logs user interface: <img src="assets/logs_ui.png" width=600></img>