Documentation

Documentation for romp: getting started, command reference, configuration, the goal contract, outcomes, architecture, and safety.

Documentation

romp is an opinionated runner for the coding agent you already have. It makes a handful of calls on your behalf — a goal contract instead of a task description, verification romp runs itself rather than trusting the agent, a review gate on every diff, and a hard stop before merge — and it does not offer a switch to unmake them. The pages below describe those choices and how to work with them.

  • Getting Started — Requirements, installation, and your first labelled issue
  • Commands — Reference for every command and flag
  • Configuration — romp.toml, config layering, and defaults
  • Goal Contract — The prompt romp hands your agent
  • Outcomes — The terminal state of every job
  • Architecture — How romp works under the hood
  • Safety — What to know before you point it at a repo