h4tcher-capability-refactor

Safe refactoring requires a cross-reference scrubbing workflow that scans code, tests, and governance files to discover impacted surfaces, rewrites delegated bodies, and produces a rename-map deliverable. This prevents broken registry entries and orchestration pipelines by verifying orphan-hits before committing.

A rename-map deliverable is an inventory of proposed changes generated during an artifact restructure, allowing maintainers to review diffs and confirm inventory or regression gates before any commits are applied to the codebase.

Managing complex refactoring involves structured tiered triage to determine scope and parallelism, followed by parallel surface scanning across canonical artifacts. Sub-agent dispatch handles delegated body rewrites while governance gates enforce compliance.

Yes, the refactoring workflow enforces a stop-before-commit safety mechanism. It applies preflight checks, cross-reference discovery, and governance gates, then halts so maintainers can review diffs and verify inventory regeneration.

Tiered triage is needed when restructuring high-impact artifacts like agents, pipelines, or governance files. It categorizes refactoring scope into levels to determine whether parallelism, sub-agent dispatch, or audit flags are required.

Automated cross-reference scrubbing requires manual governance gates and stops before commits, meaning it cannot autonomously finalize changes. Maintainers must manually review diff outputs and verify orphan-hits to ensure no references are missed.