Skills.sh is an open ecosystem and registry for “agent skills”—modular, reusable capability packages that extend what AI coding agents can do. It is developed as part of an open standard approach to making AI agents more composable and tool-driven, allowing developers to install, share, and run skills that add structured workflows and procedural knowledge to agents.

At its core, Skills.sh separates reasoning from execution by giving AI agents access to explicit, versioned commands called skills. Each skill defines a specific capability (for example: code review, SEO audit, framework best practices, or automation workflows) that an agent can invoke when relevant.

Skills are distributed through a simple CLI interface, typically installed via commands like npx skills add <repo>. Once installed, they become available for supported agents such as Claude Code, Cursor, and other AI development environments.

Each skill is structured as a self-contained package, often including metadata, instructions, scripts, and optional resources. This makes skills portable across different AI systems while remaining human-readable and easy to audit.

The platform also includes a public leaderboard and discovery system where users can browse trending and commonly used skills. Usage data is aggregated anonymously to surface the most widely adopted capabilities in the ecosystem.

Skills.sh is positioned as a “plugin layer for AI agents,” enabling teams to standardize workflows, share domain expertise, and reduce prompt complexity by loading structured, reusable behavior instead of re-describing tasks repeatedly.

Key features include:

• CLI-based installation and management of agent skills
• Modular, reusable capability system for AI agents
• Works with tools like Claude Code, Cursor, and others
• Skills defined as structured, versioned packages (SKILL.md + scripts)
• Public registry with discovery and leaderboard system
• On-demand loading to reduce context usage
• Community-driven ecosystem of reusable workflows
• Open standard for agent capability sharing

Common use cases include:

• Adding best-practice workflows to coding agents
• Automating development tasks (code review, testing, SEO audits)
• Sharing reusable engineering or domain expertise
• Standardizing AI behavior across teams
• Extending AI agents without modifying core models
• Building composable automation pipelines for developer tools

Skills.sh is developed by Vercel Labs and represents a growing approach to making AI agents more modular, deterministic, and ecosystem-driven rather than purely prompt-based.