# Skill Creator > Expertise in designing and documenting portable AI capabilities following the AgentSkills.io standard. Use this when the user wants to teach you a new workflow, standard, or technical domain. Use when this capability is needed. - Skill: `tomevault-io/skill-creator-47` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/skill-creator-47` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/skill-creator-47/raw - Safety review: pending (external: skill-scanner PASS, skillspector CAUTION) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Productivity - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/skill-creator-47 --- # Skill Creator _Expertise in designing and documenting portable AI capabilities following the AgentSkills.io standard._ You use this skill to extend the capabilities of the Vybz Squad by architecting new, reusable "Skills." A skill is a self-contained directory that encapsulates knowledge, behavioral rules, and optional executable resources. ## 1. Skill Architecture When designing a new skill, prioritize **Atomic Portability**. * **Single Responsibility:** A skill should do one thing well (e.g., `pytest-mastery`, not `python-testing-and-deployment`). * **Progressive Disclosure:** Keep the main `SKILL.md` focused on instructions. Move deep technical references to the `references/` directory and executable logic to the `scripts/` directory. ## 2. The SKILL.md Format Every skill must contain a `SKILL.md` file at its root with valid YAML frontmatter. ### Frontmatter Requirements * **`name`**: Must be lowercase, kebab-case, and match the parent directory name exactly. * **`description`**: A concise summary (max 1024 chars) explaining what the skill does and when you should use it. * **`metadata`**: Always include `author` (your persona name) and `status` (e.g., Draft, Completed). ### Body Content The body should be directive. Use the following structure: * **Introduction:** A brief italicized summary of the skill's purpose. * **Knowledge:** Facts, standards, and conceptual frameworks the agent must know. * **Abilities:** Specific actions, logic patterns, or "If-This-Then-That" behavioral rules the agent must follow. ## 3. Resource Discovery & Aggregation Vybz recursively crawls all `.md` files in a skill's subdirectories and appends them to the system prompt. * **Scripts:** List available tools in a `scripts/` directory. * **References:** Store complex schemas, man pages, or API documentation in a `references/` directory. * **Aggregation:** When you define a skill, assume the contents of these sub-files are already part of your mental model. ## 4. Vybz Workbench Conventions * **Root Path:** Internal skills reside in `src/vybz/skills/`. * **Naming:** Use descriptive, hyphenated names (e.g., `text-processing-awk--sed`). * **Output Format:** To ensure the user can immediately persist the skill using the `/save` command, you must output the file content within a code block including a filename comment. ### Example Output Pattern ```markdown # filename: src/vybz/skills/my-new-skill/SKILL.md --- name: my-new-skill description: ... metadata: author: Your Persona status: Draft --- # My New Skill ... ``` ## 5. Validation Checklist Before outputting a new skill, verify: 1. Does the directory name match the YAML `name`? 2. Is the YAML valid (check colons and indentation)? 3. Are the instructions written in an actionable, directive tone? 4. Is the `# filename:` comment present at the top of the block? --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/nick-orton) — claim your Tome and manage your conversions.