An A2A Protocol client TUI and CLI.
Install with uv:
uv tool install a2a-handlerOr run Handler in an ephemeral environment:
uvx --from a2a-handler handlerFor a full list of commands and options:
handler --helpHandler includes a reference implementation of an A2A server agent built with Google ADK and LiteLLM, connecting to Ollama for local inference.
Use any model from Ollama's library:
handler server agent --model llama3.2:1bHandler will prompt to pull missing models automatically.
Run with authentication:
handler server agent --authCopy the API key from the terminal output.
Run Handler's webhook server to receive asynchronous push notifications from A2A server agents:
handler server pushThen send messages with push notifications enabled:
handler message send http://localhost:8000 "hello" --push-url http://localhost:9000/webhookSend a message and get a response:
handler message send http://localhost:8000 "What can you help me with?"Stream responses in real-time:
handler message stream http://localhost:8000 "Tell me a story"Continue a conversation using saved session state:
handler message send http://localhost:8000 "Follow up question" --continueSave credentials for an agent:
handler auth set http://localhost:8000 --api-key my-secret
handler auth set https://api.example.com --bearer eyJhbG...Launch the terminal user interface:
handler tuiFetch an agent's card:
handler card get http://localhost:8000And validate an agent card from a URL or a file:
handler card validate http://localhost:8000
handler card validate ./agent-card.jsonUpgrade Handler to the latest version from PyPI:
uv tool upgrade a2a-handlerSee CONTRIBUTING.md.
