skills.sh

What is skills.sh?

skills.sh is "The Agent Skills Directory" for discovering and installing reusable agent skills—self-contained behavior modules that encode workflows, best practices, and domain expertise for AI agents. Instead of repeatedly prompt-engineering the same behavior, developers browse skills in the directory and attach them to compatible agent frameworks so agents can load the right procedural knowledge when needed. By treating skills like packages and the directory like an npm-style registry, skills.sh helps teams share, reuse, and standardize how their agents act across projects.

Key Features

• Central Skills Directory

  Browse a curated directory of open-source agent skills, organized with leaderboards and trending views so developers can quickly find high-impact behaviors for coding, design, marketing, automation, and more.

• Reusable Behavior Modules

  Each listed skill encapsulates a specific capability or workflow—such as React best practices, systematic debugging, SEO auditing, or AI video generation—so agents can apply consistent procedures without repeated manual prompts.

• Open Skills Format Compatibility

  Skills follow the SKILL.md–based Agent Skills format: a lightweight directory with metadata and markdown instructions that agents can progressively load, making skills portable across compatible tools and agent runtimes.

• CLI-Friendly Installation

  Designed to work with command-line tooling, skills.sh supports installing skills with a single command (for example via a `skills` CLI or `npx skills add`), so teams can wire behaviors into their agent stacks with minimal friction.

• Open-Source Ecosystem

  The directory aggregates skills from multiple public repositories and communities, giving developers a single place to discover contributions from vendors, frameworks, and independent authors while still respecting each project’s own license.

Use Cases

• Bootstrapping New Agent Capabilities : Agent developers can browse skills.sh to quickly add capabilities like web search, file processing, UX reviews, or marketing workflows, instead of building bespoke behaviors from scratch for every project.
• Standardizing Engineering Best Practices : Engineering teams can adopt skills that encode patterns such as React, Vue, Expo, testing, or debugging best practices so coding agents behave consistently with the team’s standards across repositories.
• Packaging Internal Playbooks : Organizations can author their own skills in the standard format—capturing review checklists, deployment runbooks, or domain guidelines—and use the directory model to distribute them to internal agents.
• Integrating External Tools and APIs : Teams using tool-focused skills (for example, inference.sh collections) can teach agents how to call external CLIs, APIs, or SaaS products via documented commands and workflows embedded in each skill.
• Exploring Agent Architecture Patterns : Agent and platform builders can study popular skills as reference implementations of multi-step workflows, prompt structuring, and context management patterns when designing their own agent systems.

FAQs