Author and validate agent skill directories with metadata and structure checks.
npx skills add https://github.com/involvex/skills --skill skill-creator-involvex
Please help me install this Agent Skill. Skill: skill-creator Source: https://github.com/involvex/skills/tree/main/skills/skill-creator Command: npx skills add https://github.com/involvex/skills --skill skill-creator-involvex
This Skill provides a standardized procedure and validation tooling to author, structure, and finalize agent skill directories so they are discoverable, deterministic, and production-ready.
Use the skill-creator to scaffold a new agent skill and run its validation script with the proposed name and description.
High-intent search queries and answers about installing and using this skill.
To structure an agent skill directory for discoverability, use a flat folder hierarchy with scripts/, references/, and assets/ components. This enforces progressive disclosure so the agent can efficiently locate and load production-ready capabilities.
Metadata requirements for authoring agent skills mandate strict naming conventions, specific description lengths, and neutral tone heuristics. A bundled validation CLI checks these discovery rules and reports precise formatting errors before publication.
You validate agent skill metadata by running a deterministic validation CLI script. This script checks format, enforces forward-slash paths, verifies single-level component folders, and emits clear stdout/stderr messages to flag discovery failures.
Agent skill validation scripts fail on directory structure when component folders exceed single-level depth or use incorrect path separators. The validator enforces flat hierarchies and forward-slash paths to ensure deterministic progressive disclosure.
No specific dependencies are required to scaffold production-ready agent capabilities. The skill-creator operates independently to audit directory structure, validate metadata, and enforce naming conventions without external packages.
No, you cannot use nested folders for scripts and references in agent skills. The validator enforces a flat folder structure with single-level component directories to maintain deterministic progressive disclosure.
cuembyScaffold standardized AI agent skill directories with template files.
dvy1987Package and publish validated agent skills to skills.sh and GitHub.
i3ringitExecutes a test agent in a workspace to validate integration and basic command functionality.
compozyLoad skills without AGH-specific metadata using YAML frontmatter.
