This repository offers modular "skills" to extend AI agent capabilities, targeting developers building custom systems and users of tools like Kode CLI, Claude Code, and Cursor. It enables agents to access specialized domain knowledge on demand, enhancing adaptability by providing curated knowledge rather than explicit instructions.

How It Works

The project's core philosophy is "knowledge, not code," where skills act as expert knowledge packages that inform AI models, allowing them to deduce actions. Skills adhere to the Agent Skills Spec, featuring a SKILL.md file. The approach prioritizes "expert-only knowledge" the AI model lacks, guided by Good Skill = Expert-only Knowledge - What Claude Already Knows. This knowledge-centric design promotes agent flexibility.

Quick Start & Requirements

Installation uses plugin managers:

• Kode CLI: kode plugins install https://github.com/shareAI-lab/shareAI-skills
• Claude Code: claude plugins install https://github.com/shareAI-lab/shareAI-skills
• Cursor: Copy skills/ to the Cursor skills folder. Other agents load SKILL.md on-demand. Requires an Agent Skills Spec-compatible agent.

Highlighted Details

• Available Skills: Includes skill-judge (skill quality evaluation), media-writer (content adaptation for multiple platforms), agent-builder (AI agent design), and vibe-coding (minimal-spec development).
• Skill Creation: Offers a guide and 120-point scoring system for skill quality, assessing dimensions like Knowledge Delta, Mindset + Procedures, and Practical Usability.
• Structure: Skills reside in directories with a mandatory SKILL.md, plus optional references/, scripts/, and assets/.

Maintenance & Community

Contributions are welcomed. Related projects include Kode and the Agent Skills Spec. No specific community channels or roadmap details are provided.

Licensing & Compatibility

Licensed under Apache-2.0, permissive for commercial use. Designed for compatibility with agents supporting the Agent Skills Spec, including Kode CLI, Claude Code, and Cursor.

Limitations & Caveats

Skill effectiveness depends on the underlying AI agent and Agent Skills Spec adherence. The "knowledge, not code" paradigm may require a conceptual shift. The README lacks specific performance benchmarks or individual skill limitations.