borghei/Claude-Skills

What it solves

Most AI assistants give generic answers. This project bundles 385 ready-made skills that turn your assistant into a specialist for specific jobs — preparing a board deck, writing a PRD, auditing for SOC 2, reviewing code, or planning a marketing campaign. It saves you from writing long, detailed prompts from scratch, and many skills include Python tools that compute numbers or run checks so the results are consistent, not guessed.

How it works

Each skill is a packaged prompt (with optional Python tools and metadata) that you install into your assistant. The skills are grouped into 21 domains, from leadership and engineering to compliance and marketing. You can add the whole library or just the set you need. Installation works across many assistants: in the Claude desktop app you browse plugins; in Claude Code you use plugin marketplace commands; in other tools (Cursor, Codex, Gemini CLI, Copilot, Windsurf, Cline, Goose, Aider) you run a simple npx command to drop a skill into the right folder. Many skills follow a "ask before you write" pattern — they first confirm the two or three inputs that change the deliverable, then use built-in Python tools (906 total, using only the standard library) to compute metrics, score options, or check compliance, so the output is grounded in real data rather than model guesses.

Who it's for

Founders, product managers, engineers, compliance officers, marketers, sales and customer-success teams, and anyone who uses AI assistants for professional work. It's especially useful for people who need regulated or structured outputs — like board materials, audit gap analyses, or product specs — and want those outputs to be repeatable and trustworthy. Basic installation needs no terminal; some command-line options require Node 18+.

Highlights

• 385 skills across 21 domains, including 76 expert agents and 68 single-domain agents.
• Regulated-work focus: skills for SOC 2, ISO 27001, GDPR, EU AI Act, HIPAA, and more, with gap analyses and audit-ready output.
• 906 Python tools compute scoring, sizing, and checks using only the standard library, so they can also run in CI.
• Ask-before-write behavior: skills confirm key inputs before producing a deliverable, avoiding polished guesses.
• Personas and slash commands: combine domains into roles like startup-cto or solo-founder, or use quick commands like /prd, /rice, /retro, /sprint-plan, /security-scan, and more.
• Works with 11 AI assistants, including Claude, ChatGPT, Gemini, Cursor, Copilot, and others.
• Extensible: contributors can add new skills following the Skill Authoring Standard.