scope-control

To prevent scope creep, enforce the smallest viable change set for incoming tasks. This keeps diffs focused and reviewable by validating work size and proactively proposing decomposition for large features before implementation begins.

The best way to break down large tasks is automatic decomposition. When a task exceeds a single-session scope, the system proactively proposes splitting it into smaller steps, initiating a ralph-loop, or suggesting a worktree.

Scope monitoring detects scope growth during active work and pauses progress automatically. It then prompts a re-scope with the user to prevent runaway changes, ensuring the task remains focused and within acceptable boundaries.

Yes, you can enforce minimal changes silently by operating as a high-priority gate. This identifies scope size and suggests rescoping actions without user disruption, integrating directly with routing and chain-to skills.

You should propose task decomposition whenever a broad feature request arrives. This produces a compact set of smaller implementation tasks and recommends the appropriate execution loop or review gate to handle the work.

Yes, scope control works with routing and chain-to skills. It applies to all received tasks and gate evaluations, operating silently to validate incoming work and integrate with existing workflow routing.