borghei/Claude-Skills

Claude Skills – a reusable library of AI‑agent “skills”

What it is – An open‑source collection of 385 ready‑to‑use skills (prompt‑plus‑code bundles) that can be loaded into Claude, ChatGPT, Gemini, Copilot, Cursor and other LLM‑assistant platforms. Each skill pairs a natural‑language description with a small, pure‑Python tool (standard‑library only) and optional YAML metadata. The repo also ships 77 pre‑defined expert agents (personas such as startup-cto, growth‑marketer, cs‑fintech‑advisor) that expose groups of related skills via slash‑commands.

Why it matters – Most LLM prompt libraries stop at static text templates. Claude Skills goes a step further by providing executable Python helpers that can be dropped into CI pipelines, MCP servers, or local scripts, letting non‑engineers run real compliance checks, audit reports, sprint retrospectives, LinkedIn content generation, etc., without writing code themselves.

Key numbers (from the README)

• 385 skills across 21 domains (PM, Engineering, Marketing, Compliance, Legal, Finance, …)
• 900 pure‑Python tools (no external dependencies)
• 77 expert agents / personas
• 18 regulated‑industry compliance frameworks (SOC 2, ISO 13485, HIPAA, GDPR, EU AI Act, …)
• 11 supported AI assistant platforms

How you use it

1. Browser‑based – Visit the hosted skill library (https://borghei.github.io/Claude‑Skills), click Try it now, copy the prompt, and paste it into Claude.ai, ChatGPT, Gemini, etc.
2. Claude Desktop – In the Claude app open Customize → Browse plugins and enter borghei/Claude‑Skills. All skills become available as native plugins.
3. CLI for developers – Run npx @borghei/claude-skills add senior-fullstack. The tool detects which AI‑assistant you are using (Claude Code, Cursor, Codex, Gemini CLI, Copilot, Windsurf, Cline, Aider, Goose) and drops the appropriate skill files into the correct directory. Requires Node 18+, no Python runtime needed for the installer.
4. Bundles – Pre‑packaged skill sets (e.g., SaaS Founder Kit, Compliance Kit, PM Starter Pack) can be installed in one step; the repo’s bundles.json maps each kit to the underlying skill IDs.
5. Slash commands – Inside a supported assistant you can invoke commands like /retro, /security-scan, /tdd, /prd etc., which trigger the associated skill logic.

What you get out of the box

• Compliance workflows – Checklists, gap analyses, and audit‑ready outputs for SOC 2, ISO 13485, MDR, FDA, NIS2, DORA, HIPAA, GDPR, EU AI Act, etc.
• Project‑management tools – Sprint planning, retro facilitation, PRD generation, technical‑debt tracking, changelog creation, and scoring rubrics.
• Business‑operations helpers – Capacity planning, vendor management, research‑ops design, markdown‑to‑HTML publishing pipeline.
• LinkedIn automation suite – 12 tools for profile building, post drafting, comment replies, and engagement analysis (offline only; nothing is posted automatically).
• Multi‑agent orchestration – Four patterns (sequential pipeline, parallel fan‑out, supervisor delegation, consensus voting) defined in the repo’s Orchestration Protocol standard, enabling complex workflows that chain several skills together.
• Expert personas – Activate a persona with /persona startup-cto (or just mention it) to expose a curated set of skills for that role.

Documentation & community

• Full docs site: https://borghei.github.io/Claude‑Skills
• Skill reference, agent reference, usage guide, installation guide, standards, and contribution guidelines are all in the docs/ folder.
• Open to PRs: contributors add new skills by following the Skill Authoring Standard (YAML front‑matter + Python tool).
• License: MIT + Commons Clause (free for OSS, personal, education, internal business use; cannot be sold or repackaged as a paid product).

Who it’s for – Product teams, founders, marketers, compliance officers, and developers who want to harness LLM assistants for concrete, repeatable business processes without building the tooling from scratch.

---

All details above are taken directly from the repository’s README; no additional features have been inferred.