What problem does it solve? Agent-created and local Hermes skills accumulate quality issues over time: missing frontmatter, bloated descriptions that waste tokens in every turn, duplicated content across skills, stale unused skills, and procedural rules cluttering memory. This Skill provides a systematic audit-and-fix workflow instead of ad-hoc manual cleanup. ## Core Features & Use Cases - Inventory & Diagnosis: Cross-references hermes skills list with .usage.json to classify skills as unused, externally injected (e.g., CC Switch), or needing fixes, with a diagnostic table mapping problems to signals and fixes. - Frontmatter & Description Standardization: Patches missing license and metadata.hermes.tags, and compresses descriptions to ≤120 characters in "Use when <trigger>" format to reduce per-turn token cost. - Content Dedup & Memory Migration: Replaces duplicated prose with one-line cross-references and removes procedural rules from memory that already live in skills, targeting under 70% memory capacity. - Curator Setup & Export: Configures automated lifecycle management (curator.consolidate, stale/archive thresholds) and exports local skills as a zip for cross-machine migration. - Use Case: After noticing hermes skills list shows unexpected local skills and memory is at 94%, run a full audit to prune unused skills, fix frontmatter, deduplicate content, clean memory, and detect external tool injection. ## Quick Start Ask the agent to audit all local Hermes skills, fix frontmatter and description issues, deduplicate content, and report memory usage.
