auto

Autonomous AI coding loops automate unattended development by executing tasks defined in a prd.json file. The AI agent independently selects tasks, implements code, runs quality checks, and commits changes through a self-contained loop.

An autonomous development loop is a self-contained process where an AI agent iteratively selects and implements tasks from a PRD. It manages coding, quality checks, and commits while operating with fresh context per iteration, relying on persistent files for state continuity.

Start the autonomous coding loop by running the command `samuel auto start --yes`. This initiates the unattended workflow, allowing the AI agent to begin selecting and executing tasks defined in your project.

Yes, autonomous AI development tracks progress by documenting updates in a persistent progress.md file. Because the agent operates with fresh context per iteration, it relies on this file alongside prd.json to maintain state continuity across loops.

Yes, you can use it for complex features defined in a PRD. The autonomous coding loop breaks down the requirements, writes the code, tests it, and commits the changes, enabling away-from-keyboard development without manual intervention at each step.

You should avoid using an autonomous coding loop for tasks requiring constant human supervision or real-time contextual decisions. Since it operates unattended with fresh context per iteration, it is unsuitable for processes needing continuous manual guidance.