What problem does it solve? Teams extending the EVA agentic platform need a consistent, safe way to create, rename, refactor, or retire runtime skills without duplicating existing capabilities or bypassing tenant isolation and approval gates. ## Core Features & Use Cases - Standardized Skill Authoring: Enforces a canonical directory shape with SKILL.md frontmatter, optional scripts, references, and templates. - Safety-First Guardrails: Embeds the EVA Runtime Skill Policy, requiring org_id tenant boundaries, secret protection, and explicit approval for external writes or destructive actions. - Audit and Removal Criteria: Defines when to update, merge, or delete skills that duplicate others, leak credentials, or skip approval gates. - Use Case: When a recurring workflow emerges in EVA, use this Skill to scaffold a new runtime skill with narrow trigger wording, then verify it with the skills-runtime test suite. ## Quick Start Ask EVA to create a new runtime skill for your repeatable workflow and have it validated against the runtime skill policy.