This project was scaffolded with the AgentRegistry CLI. It gives you a working ADK Python agent wired for MCP tools and ready to publish through AgentRegistry.
- Provider: gemini
- Model: gemini-2.5-flash
Update myagent/agent.py if you need to switch providers, add tools, or
change the root instructions.
-
Install uv if you haven't already.
-
From the project root run:
uv sync uv run myagent
-
Use
arctl agent run .to launch the local chat experience with docker compose.
-
Build (and optionally push) the container image:
arctl build . --push -
Publish the agent so the registry can serve it to clients:
arctl apply -f agent.yaml
-
Share the resulting agent link or deploy it to your runtime of choice.
Need MCP servers? Edit agent.yaml manually to append MCP entries under
spec.mcpServers, then re-apply once you're happy with the configuration:
arctl apply -f agent.yaml