awesome-claude-skills: 1,000 Claude Skills, Defining Agent Behavior

Claude Skills are reusable instruction packages that define how AI agents perform specific tasks. Each skill is provided as a folder containing YAML metadata, Markdown descriptions, and scripts. Introduced by Anthropic in October 2025 and released as an open standard in December, they are currently supported by various coding agents such as Claude Code, Codex, Cursor, and Gemini CLI.

This repository is a curated list of over 1,000 production-grade skills and plugins organized by category. It covers a wide range of areas including document processing, development tools, data analysis, and business automation. Each skill provides the workflows and guardrails agents need when performing real tasks. Designed to go beyond simple text generation, they enable integration with external services such as sending emails, creating issues, and posting to Slack.

At the start of a session, only the name and description of a skill are loaded to minimize the context window. The full body is loaded only when the agent determines the skill is relevant to the current task, and auxiliary files are loaded on demand. This allows a single agent to hold hundreds of skills without wasting tokens. Unlike MCP servers or individual tools, skills are a layer that defines the sequence of actions and rules following connections and tool calls.

By combining with the Composio MCP Gateway, authentication, access control, and audit logs for over 1,000 app integrations can be handled through a single endpoint. Installing the connect-apps plugin allows you to configure Claude to perform actual tasks across various services such as Gmail, Slack, GitHub, and Notion. It is valuable in any environment where developers to business users want to delegate repetitive digital tasks to agents.