frankxai/claude-skills-library

---

[!TIP] New — curated lanes. The best of this library, curated and actively maintained as two installable flagships: frankxai/skills for AI architects (MCP, orchestration, model routing, context engineering) and frankxai/creator-skills for creators (video routing, music, images, brand voice). Install with npx skills add frankxai/skills or npx skills add frankxai/creator-skills. This library remains the full catalog.

Join the skills community

• Ask a question, share a working example, or suggest an idea. The discussion space is open; start the conversation that would help you build.
• Report a bug or request a skill with the affected skill and a small, shareable example.
• Contribute a fix or skill. We welcome evidence from real use and clear source attribution.
• Report a vulnerability privately. Keep exploit details out of public threads.

See the support guide for routing and the code of conduct for participation expectations. Remove credentials and private material before posting.

⚡ Quick start

Option A — install as a Claude Code plugin (recommended). One command, auto-updating:

/plugin marketplace add frankxai/claude-skills-library
/plugin install claude-skills-library@claude-skills-library

Skills are now discovered automatically and activate when relevant.

Option B — clone and copy the skills you want:

git clone https://github.com/frankxai/claude-skills-library.git ~/claude-skills-library
mkdir -p ~/.claude/skills

# one skill…
cp -r ~/claude-skills-library/free-skills/mcp-architecture ~/.claude/skills/
# …or all of them
cp -r ~/claude-skills-library/free-skills/* ~/.claude/skills/

Some skills are namespaced (e.g. free-skills/anthropic/pdf). Copy the folder that directly contains the SKILL.md you want. Restart Claude Code and the skills are discovered.

Then just work. Skills trigger on their own, or invoke one explicitly:

"Use the greek-philosopher skill to examine my career decision."
"Design an MCP server with the mcp-architecture skill."
"Review this PR with the github-code-review skill."
"Write a Suno prompt for a cinematic trailer track."

---

🚀 What is this?

A curated, open-source library of Agent Skills — the portable, model-invoked capability format from Anthropic. Drop a skill into your agent and it gains expert knowledge in a domain: it loads automatically when relevant and stays out of the way when it isn’t.

Every skill here:

• ✅ Is a self-contained folder with a SKILL.md and optional references/, scripts/, assets/.
• ✅ Has spec-compliant frontmatter (name + description) — enforced by scripts/validate_skills.py.
• ✅ Is progressive-disclosure friendly: a tight description for routing, deep guidance loaded only on use.
• ✅ Carries a trigger-rich description so the right skill activates from a library this large.

114 skills across AI agents, MCP & SDKs, frontend, Oracle/cloud, content & creative production, engineering workflow, and personal performance. All free. MIT-licensed.

---

🗂️ Skill categories

Counts below are the source-of-truth groupings from docs/CATALOG.md, regenerated from the repo by scripts/generate_catalog.py.

See the full catalog for every skill with its description.

---

⭐ A few to start with

---

🆚 Skills vs. MCP vs. prompts

They solve different problems and compose well together.

Use a skill to teach the agent how to do something well; use MCP to give it access to a system; use a prompt for a one-off. Many skills here pair naturally with an MCP server.

---

🌐 Works with six runtimes

Agent Skills are a portable format. This library documents an import path for each major agentic runtime, with one adapter guide per runtime in runtimes/:

See MULTI_RUNTIME.md for the universal-MCP rationale and the full strategy.

---

🧱 Anatomy of a skill

free-skills//
├── SKILL.md          # required — frontmatter + instructions
├── references/       # optional — deep docs loaded on demand
├── scripts/          # optional — executable helpers the agent runs
└── assets/           # optional — templates, fonts, icons used in output

Every SKILL.md opens with two-field frontmatter:

---
name: mcp-architecture          # lowercase, hyphenated, ≤ 64 chars
description: Design and implement MCP servers... Use when architecting an MCP server...
---

The description is the most important line you’ll write — it’s what the model uses to decide when to load the skill. State what it does and when to use it, with trigger keywords.

📐 Authoring standard: spec/README.md · Start here: template/SKILL.md

---

✅ Quality & validation

Three zero-dependency scripts enforce the standard across every SKILL.md:

python3 scripts/validate_skills.py          # frontmatter + structure (exits non-zero on failure)
python3 scripts/generate_catalog.py         # regenerate docs/CATALOG.md
python3 scripts/generate_catalog.py --check  # CI-friendly drift check
python3 scripts/check_links.py              # verify no broken internal links

All three run automatically on every push and pull request via GitHub Actions. validate_skills.py checks that every skill has well-formed UTF-8 frontmatter, a spec-valid name, a non-empty description (≤ 1024 chars), and the exact filename SKILL.md; it also warns (without failing) on bodies over 500 lines and on reserved-word names. Run it before opening a PR.

A browsable catalog is generated by scripts/generate_site.py into docs/index.html for serving via GitHub Pages.

---

🤝 Contributing

New skills, deeper examples, fixes, and better docs are all welcome.

1. Read CONTRIBUTING.md and the authoring standard.
2. Copy template/ to free-skills// and write your SKILL.md.
3. Run python3 scripts/validate_skills.py and python3 scripts/generate_catalog.py until both are clean.
4. Open a pull request.

---

🔗 Related repositories

This library is the public, install-from storefront. Two sibling repos sit behind it:

• frankxai/agentic-creator-os — the canonical, larger internal source for skills, agents, and commands. New skills are authored and battle-tested in ACOS first; the strongest, most broadly useful ones are mirrored out to this public library.
• frankxai/Starlight-Intelligence-System — the substrate: the persistent context, memory, and multi-agent layer the skills run on top of.

In short: Starlight is the substrate, ACOS is the canonical skill source, and this repo is the free public catalog.

---

❓ FAQ

Does loading 114 skills bloat my context? No. Only each skill’s name + description is preloaded for routing; the full body loads only when a skill triggers, and references/ load only when read. That’s the whole point of progressive disclosure.

How do skills get picked from a library this large? By the description. Every skill here states what it does and when to use it with explicit trigger keywords, so the model selects the right one. If a skill mis-triggers, sharpen its description — PRs welcome.

Do I have to install all of them? No. Install the plugin for everything, or copy just the folders you want into ~/.claude/skills/.

Is this affiliated with Anthropic? No. It’s an independent, MIT-licensed community library built on Anthropic’s open Agent Skills format. Some skills under free-skills/anthropic/ are upstream reference skills that retain their original licenses.

Can I use these outside Claude Code? Yes — see runtimes/ for Antigravity, OpenCode, Codex, Gemini CLI, and Cursor.

---

Would you pay for a skill that proves it works?

Every skill here is free, MIT, and yours. That is not changing.

But a skill is prose until something tests it. All 103 in this library read well. None of them ships evidence that it still behaves under pressure — and that is exactly how a skill quietly degrades an agent instead of improving it, in a way you notice three sessions later and blame on the model.

There is no paid product here today. Before building one, the honest move is to find out whether it should exist. The idea on the table: a pack where every skill ships with its promptfoo eval suite, the cases it passes, and — the part nobody publishes — the cases it fails.

If that is worth money to you, say so and name the band. If it is not, say that instead; it is the more useful answer, and it is the one that stops this from being built.

Tell us what it is worth →

The first fifty people to answer get one skill with its full eval suite and run output, free, whatever happens next. At fifty answers the eval harness is published open-source regardless of whether the pack is ever built.

---

📜 License

MIT — free to use, modify, and distribute. See LICENSE. Skills under free-skills/anthropic/ retain their original upstream licenses (check the folder for a LICENSE.txt).

---