classifying-actions

To classify mutating actions for auto-execution or architect approval, provide an action_id and repository root to apply a 5-step triage rule and decision matrix, returning an effective autonomy class of A, R, or N.

Autonomy overrides for classifying board-superpowers actions use a dual-layer system where project-layer overrides take precedence over user-layer overrides during conflicts to compute the final action class.

When parsing fails during a mutating-action triage decision, the classification applies a conservative fallback to ensure safe governance, defaulting to a restrictive autonomy class rather than auto-executing.

To start triaging mutating actions for governance decisions, you need to provide a valid action_id and a repository root path to consult the action catalogs and derive the final class.

Yes, you can escalate Auto actions to require architect approval by applying the 5-step triage rule, which evaluates specific conditions to override the default decision matrix classification.

The decision matrix serves as a rulebook authority for governing mutating actions, using matrix.md and action-id catalogs to derive the final autonomy class for safe and auditable repository changes.