What problem does it solve? Technical debt findings from audits and code reviews get lost or duplicated without a central tracking system. This Skill maintains a single persistent backlog file (memory/backlog.md) where issues are recorded, deduplicated by fingerprint, prioritized, and resolved over time. ## Core Features & Use Cases - Backlog CRUD with deduplication: Add, list, fix, wontfix, delete, archive, and re-open items using fingerprint-based dedup keys so regressions are re-opened under the same ID instead of filed as new work. - Prioritization scoring: Ranks OPEN items with a (Impact + Risk) x (6 - Effort) formula, optionally enhanced by CodeSift reference counting to measure blast radius. - Batch suggestions: Groups items by pattern (same audit gate, same test pattern, same module) and proposes batch fix commands like zuvo:fix-tests or zuvo:refactor. - Use Case: After running a code audit that finds 12 issues, persist them to the backlog, run prioritize to rank them by urgency, then use suggest to get a batch fix plan for the 5 items sharing the same failure pattern. ## Quick Start Ask the AI to list the current tech debt backlog and add a new high-severity item describing a race condition in payout.service.ts.