Here's a quick guide to your agent project (opens in a new tab).
The starting codebase is small but very extensible.
- It's built atop the Steamship SDK (opens in a new tab), which you can think of as Flask for AI Agents.
- It deploys to the Steamship Cloud (opens in a new tab), which manages all the AI orchestration & monitoring to power your agent.
There are only three critical files required to run and deploy an agent to Steamship:
steamship.jsoncontains information used during deployment: your agent's name, version, etc.
requirements.txtcontains pip modules your agent depends upon. This is necessary to run in the cloud.
AgentService, which is a Flask-like API container
Agentwithin it, with access to Tools, LLMs, Memory, and async task queues
Everything else in the starter project is just a convenience file to help things build & run smoothly.