What problem does it solve?
Managing XGuard content-scanning policies requires calling the orchestrator's admin endpoints by hand, which is error-prone and risky when destructive changes affect every subsequent scan. This Skill provides a safe CLI wrapper for inspecting, editing, backing up, and restoring per-label policy options.
Core Features & Use Cases
- Policy Inspection: Fetch current or default policy options for text or prompt scan modes via GET requests.
- Guarded Writes: PUT, reset, and import operations require an explicit --writable flag, preventing accidental global policy changes.
- Backup and Restore: Export the full policy registry to JSON before edits and import it back to roll back.
- Use Case: During scanner prompt tuning, export a backup, pull the current prompt-mode policies, edit the policy text, ship the new version, then track FP/FN rates against the new policyHash in the audit log.
Quick Start
Ask the AI to show the current XGuard prompt-mode policy options using the xguard-manager skill.