alirezarezvani/claude-skills

⭐ 14,298  ·  Python  ·  GitHub Repo

232+ Claude Code skills & agent plugins for Claude Code, Codex, Gemini CLI, Cursor, and 8 more coding agents — engineering, marketing, product, compliance, C-level advisory.

agent-plugins agent-skills agentic-ai ai-coding-agent anthropic-claude claude-ai claude-code claude-code-plugins

1-Sentence Summary

235 production-ready skill packages that turn any coding agent into a domain expert.

🔥 Key Capabilities & USP

• 235 Skills Across 9 Domains: Engineering, marketing, product, compliance, finance, project management, business growth, regulatory/quality management, and C-level advisory. Pain point solved: AI agents lack domain-specific knowledge out of the box—this fills that gap with structured, reusable expertise.
• Multi-Tool Compatibility: A single convert.sh script transforms all skills into native formats for 12 tools—Claude Code, OpenAI Codex, Gemini CLI, Cursor, Aider, Windsurf, Kilo Code, OpenCode, Augment, Antigravity, and more. Pain point solved: No vendor lock-in; invest in skills, not tools.
• Personas System: Pre-configured agent identities (Startup CTO, Growth Marketer, Solo Founder) with curated skill loadouts and distinct communication styles. Pain point solved: Teams need different perspectives for different tasks, not a one-size-fits-all agent.
• Orchestration Protocol: Four patterns (Solo Sprint, Domain Deep-Dive, Multi-Agent, etc.) for coordinating personas and skills across domain boundaries. Pain point solved: Complex projects require cross-domain collaboration that single agents can't handle.
• Zero-Dependency Python Tools: All 305 CLI scripts use only Python stdlib—no pip install, no dependency hell. Pain point solved: Immediate execution without environment setup friction.

USP: The most comprehensive open-source skill library that works across 12 different AI coding tools with zero external dependencies.

Technical Architecture

Component	Technology	Purpose
Skill Packages	Markdown (SKILL.md) + Python stdlib scripts	Structured instructions and executable tools per domain
Conversion Engine	Bash (scripts/convert.sh)	Transforms skills into native formats for Cursor (.mdc), Aider (CONVENTIONS.md), Windsurf, etc.
Plugin System	Native marketplace commands + manual copy	Installation via Claude Code /plugin, Gemini CLI scripts, or direct folder placement
Personas	Markdown agent definitions	Pre-configured identities that load specific skill sets
Orchestration	Protocol patterns (4 types)	Coordinates multiple agents/personas across domain boundaries
Three-Layer Model	Skills → Agents → Personas	Skills define how, agents define what, personas define who

Quick Start Guide

Install for Gemini CLI:

git clone https://github.com/alirezarezvani/claude-skills.git
cd claude-skills
./scripts/gemini-install.sh
> activate_skill(name="senior-architect")

Install for Claude Code:

/plugin marketplace add alirezarezvani/claude-skills
/plugin install engineering-skills@claude-code-skills

Install for OpenAI Codex:

npx agent-skills-cli add alirezarezvani/claude-skills --agent codex

Install for OpenClaw:

bash <(curl -s https://raw.githubusercontent.com/alirezarezvani/claude-skills/main/scripts/openclaw-install.sh)

Convert for all tools:

./scripts/convert.sh --tool all
./scripts/install.sh --tool cursor --target /path/to/project
./scripts/install.sh --tool aider --target . --force

Verify installation (Cursor example):

find .cursor/rules -name "*.mdc" | wc -l

Activate a persona (Claude Code):

cp agents/personas/startup-cto.md ~/.claude/agents/

Pros, Cons & Use Cases

Pros

• Extremely comprehensive: 235 skills across 9 domains—covers engineering, marketing, compliance, finance, and C-level advisory.
• Zero dependency overhead: All 305 Python scripts use only stdlib. No requirements.txt, no virtualenv, no npm install.
• Multi-tool by design: One skill library, 12+ supported agents. Conversion is a single command.
• Modular and selective: Install only the domains you need. No bloat.
• Open source (MIT): Free to use, modify, and extend. Active community with 14,298 stars.

Cons

• Manual conversion for some tools: Not all 12 tools have one-click install; some require running convert.sh and copying files.
• Skills are instruction packages, not autonomous agents: They provide structured guidance and scripts, but don't execute independently.
• Orchestration learning curve: Cross-domain work requires understanding the four orchestration patterns, which adds complexity for simple use cases.

Who should NOT use this?

• Users of a single AI coding tool who want deep, tool-specific integrations: If you only use Cursor and want native Cursor-only plugins, this project's multi-tool abstraction may feel like unnecessary overhead.
• Teams that need fully autonomous, self-executing agents: These skills are structured instructions and helper scripts—they don't replace an autonomous agent framework.
• Developers who prefer GUI-based plugin management: Installation for some tools still requires terminal commands and manual file copying.

Ideal Use Cases

• Startup CTOs who need their AI agent to switch between engineering, product strategy, and compliance advisory in a single session.
• DevOps teams standardizing agent behavior across a mixed tool ecosystem (Claude Code, Cursor, Aider, Windsurf).
• Freelance developers who want one skill library that works across all their client projects regardless of the AI tool in use.
• Enterprise compliance teams who need structured, repeatable processes for regulatory tasks embedded directly into their coding agents.

Community & Activity

This project is on fire. With 14,298 stars and active maintenance (last updated May 2026), it's rapidly becoming the standard for cross-platform AI agent skills. The community has contributed skills across 9 domains, and the project maintainers are clearly committed to keeping pace with the rapidly evolving AI coding agent landscape. The 235+ skills and 12-tool support demonstrate serious momentum—this isn't a side project, it's an ecosystem in the making.