What problem does it solve? After a security or code audit produces a remediation plan, teams still face the risky work of actually applying fixes in the right order, with approvals, validation, and rollback safety. This Skill executes an approved audit remediation plan deterministically, preserving finding traceability, dependency order, and durable progress state so partial work is never marked complete. ## Core Features & Use Cases - Scoped execution modes: Run all eligible items, a single phase (milestone), a single finding (AUD-id), or resume from the last valid checkpoint. - Governed execution: Policy evaluation before every gated item, phase-level approval gates, and separate point-of-action approval for destructive, external, or privileged actions. - Validation and rollback: Each item runs the narrowest falsifying validation, follows plan rollback guidance on failure, and blocks dependent items until prerequisites pass. - Event-sourced progress: Appends outcomes to an immutable execution log, creates content-addressed checkpoints, and updates project handoff artifacts after every phase. - Use Case: After audit-plan-remediation produces reports/audit-remediation-plan.json, run /audit-remediation -Phase containment to execute the containment milestone and its prerequisites with approvals, validation, and checkpointed handoff. ## Quick Start Ask the agent to run /audit-remediation -All to execute every eligible item from the approved remediation plan in dependency order with phase approvals.
