OpenRig
← All examples

Keep each task tied to the goal

The developer asks for an app settings screen and tells dev-ui@factory to check the current milestone before adding more. The immediate task is small; the important question is how much of the larger product it should cover.

What happens

  1. Read the mission

    dev-ui@factory uses rig scope to inspect first-release. The mission’s intent is to let customers try the new feature this week.

  2. Read the slice

    The settings slice narrows that goal: provide the settings needed for the trial, with administration tools left for a later milestone. Both pieces of intent were written into the project workspace.

  3. Apply the boundary

    The agent records the decision in a project note and commits to building only the trial settings. Its answer connects today’s work to the larger goal instead of expanding the task into a full administration system.

What OpenRig does

OpenRig scopes give the agent a way to inspect the mission and slice associated with the work. Their files carry the purpose and boundaries the team has authored.

What Claude or Codex does

The coding agent reads that intent, interprets what belongs in the milestone and chooses its next action. OpenRig does not decide the product scope for it. Writing notes/settings.md is an ordinary shell action.

When this helps

This is useful when a plausible extra feature would take the team away from the outcome the developer actually needs. The agent can check the purpose of its task before spending time on the wrong expansion.