# Skills Catalog — Template Exact section structure every `-Skills-Catalog.md` must follow. Read during Step 6 of the parent SKILL.md. --- ## File header ```markdown # Skills Catalog > **Rated, persona-sorted catalog of skills / extensions / plugins / rules / MCP servers.** > Last updated: . Research window: → . > Companion files: [the five sibling files] --- ``` --- ## Required sections (in order) ### 1. Primer — ``'s extensibility surface Table mapping primitive → file/location → what it does. Example primitives: - Skills (SKILL.md directories) - Plugins (native code bundles) - Rules (e.g., Cursor `.mdc`, Gemini `GEMINI.md`, Codex `AGENTS.md`) - Memory files (Claude `CLAUDE.md`, Codex `AGENTS.md`, OpenClaw `MEMORY.md` / `SOUL.md` / `HEARTBEAT.md`) - MCP servers - Gems (Gemini) / Notepads (Cursor) / Artifacts (Claude) - Hooks Also a one-paragraph note on `agentskills.io` cross-tool compatibility. ### 2. Install & management commands Copy-pasteable. Cover: install, search, list, update, disable, remove, publish. Every command must actually exist — don't invent flags. ### 3. Rating legend Either copy the legend from [`rating-system.md`](./rating-system.md) inline, or link to it. Inline is preferred for standalone readability. ### 4. Built-in / official foundation skills Skills that ship with the tool or are maintained by the vendor's own GitHub org. All should be ✅ Verified by definition. ### 5. Marketplaces / meta-collections Curated lists, awesome-lists, registries, meta-skills. Rate each as a collection (not its individual contents). ### 6. Skills by persona Sub-sections 6.1 through 6.8, in the same persona order as `personas.md`: - 6.1 PhD / research students - 6.2 Solopreneur / startup founder - 6.3 Marketer / growth / agency - 6.4 Designer - 6.5 Video / podcast / creator - 6.6 Software developer - 6.7 Personal productivity / PKM / student - 6.8 Sales / finance / ops Each sub-section: a table with columns `Skill | URL | Rating (validity + stars) | Role`. Minimum 5 entries per persona if meaningful coverage exists; fewer is acceptable with an honest note. ### 7. Cross-tool skills Skills that honor `agentskills.io` and work across ≥ 2 of the five tools in this series. These are disproportionately valuable — flag them specifically. ### 8. Unusual / frontier skills Mirror section 12 of the playbook but with install commands. Show the tail of the skill distribution. ### 9. Starter kits by persona Copy-paste-able install blocks. Example: ````markdown ### Developer starter ```bash install install ... ``` ```` One starter kit per persona that has meaningful coverage. Plus one "safety-first starter" (install these BEFORE any other skills). ### 10. Creating your own skill Minimal viable SKILL.md example for this tool (YAML frontmatter shape, trigger terms, publishing command). Link to the vendor's official skill-authoring doc. Include meta-skills for building skills if any exist (e.g., Anthropic's `skill-creator`, OpenClaw's `agentskills-io`). ### 11. Safety / vetting protocol Specific to tools with system access (all five to varying degrees). Include: - Install-gate skills (e.g., `azhua-skill-vetter` for OpenClaw) - Sandbox options - Audit logging options - Cost monitoring - Prompt-injection screening ### 12. Tier-1 must-installs A single table, 5–8 items max, with the rationale for each. This is the section most users will read. Be ruthless — no sympathy picks. ### 13. Cross-catalog navigation Links to the four sibling catalogs with a one-line description each. Reminds the user that `agentskills.io`-compliant skills are portable. ### 14. Master source index Same groupings as the playbook's section 16: Official, Curation, How-to, Context/culture. ### 15. (Optional) Appendix A — Ratings summary Appendix table of all skills in the catalog with their ratings, useful for quick scanning. Optional if ratings are already inline everywhere. --- ## Table formatting contract Every skill table has at minimum: | Column | Required | Example | |---|---|---| | **Skill** | yes | `steipete/github` | | **URL** | yes | clickable Markdown link to canonical source | | **Rating** | yes | `✅ ★★★★★` or split into two columns | | **Role / Why** | yes | 6–12 words, not a sentence | Do not add columns that would be empty for most rows. --- ## Diff-friendly rules - Alphabetical order within each persona table (or by descending usefulness rating — pick one convention per file and stick with it) - Heading text identical across runs - Every skill entry one row, one line (no multi-line rows) - Ratings as literal characters `✅ 🟢 🟡 🔴 ★` — not images or HTML - Dates always `YYYY-MM-DD` --- ## Length guide | Section | Target lines | |---|---| | Header + sections 1–3 | 40–80 | | Section 4 (foundation) | 20–60 | | Section 5 (marketplaces) | 20–50 | | Section 6 (personas, all 8) | 150–300 | | Sections 7–11 | 30–80 each | | Section 12 (tier-1) | 15–30 | | Sections 13–14 | 30–60 | | Optional appendix | 30–80 | | **Total body** | **400–800 lines** | Skills catalogs can justifiably be longer than playbooks because they're reference material. Users read them in chunks.