Your first project
Go from zero to a running agent in four steps. Have the CLI installed and Docker ready.
Create an agent
Create a new agent harness:
The wizard asks for a name, LLM provider, and agent interface. Pass --yes to accept defaults, or --model to skip the provider step:
Build your agent logic
The agent harness includes starter source code, astropods.yml, and two files that describe the project to coding agents:
AGENTS.md: explains the project structure, the spec format, and the conventions for writing a proper Astropods agent. Any coding agent (Claude Code, Copilot, Cursor, etc.) can read this to understand how to implement and extend your agent correctly.CLAUDE.md: Claude Code-specific instructions for working in this project.
You can write your agent logic manually, or open the project in a coding agent and let it do the heavy lifting:
Tell it what your agent should do and it will implement the logic, update the spec, and add any required inputs.
Next steps
- Your first blueprint: build and push your agent to the registry
- Deploy your first agent: run a blueprint as a live hosted agent
- Managing secrets: store API keys in the account vault
Try a pre-built agent
hello-astro is the simplest possible Astropods agent (no configuration required). Use it to verify your local setup and as a starting point for understanding the project structure.
Once it’s running locally you can push it to the registry and deploy it as a live agent: