# Awesome Claude Skills — 1,000+ Reusable AI Agent Workflows

> Composio's curated directory now hosts over a thousand Claude Skills spanning document processing, code automation, security auditing, CRM workflows, and more — all built on Anthropic's open SKILL.md standard.

_Source: ComposioHQ via GitHub · 2026-07-23 · 4 min read · Verified against primary sources_

Canonical: https://iyu.app/e/awesome-claude-skills

## The 60-second version

Composio's Awesome Claude Skills repository has curated over 1,000 reusable AI agent skill packages — from document processing to CRM automation — all built on Anthropic's open SKILL.md standard.

**Key points**

- The SKILL.md format is an open standard for teaching AI agents how to perform specific tasks, adopted by Claude Code, Codex, Cursor, Gemini CLI, and more.
- Skills load progressively (~100 tokens per skill at session start), allowing a single agent to carry hundreds without bloating context.
- Composio contributes pre-built workflow skills for 78 SaaS apps including Salesforce, Jira, Slack, GitHub, and Stripe.
- The directory spans 15+ categories: document processing, development, data analysis, security, CRM, email, social media, e-commerce, and more.

**Verdict.** For anyone building with AI coding agents, this is the single best resource to discover what skills exist and what's possible — and the open format means your own skills are equally portable.

## Full explainer


### The Skills Revolution — Why AI Agents Need Instructions, Not Just Tools

Since Anthropic introduced the **Skills** format in October 2025 — and open-sourced the standard three months later — the ecosystem has exploded. Skills are reusable instruction packages that teach an AI agent *how* to complete a specific class of task, packaged in a simple folder with a `SKILL.md` file at its heart. Unlike MCP servers (which define *connections to external systems*) or tools (which define *individual functions*), skills define **the workflow itself**: what to do, in what order, and what to watch out for.

The key innovation is progressive loading. When an agent session starts, it sees only each skill's name and description — roughly 100 tokens each. The full instructions load only when the agent decides the skill is relevant. This is how a single agent can carry hundreds of skills without bloating its context window.

ComposioHQ's [Awesome Claude Skills](https://github.com/ComposioHQ/awesome-claude-skills) repository has become the de facto community hub, curating over 1,000 skills across dozens of categories — and it's completely free.


### Catalog — Inside the Directory — What's Available


#### Document Processing

Skills for working with Word docs (tracked changes, formatting), PDFs (extraction, annotation, merging), PowerPoint slides (generation, layout), and Excel spreadsheets (formulas, charts, pivot tables). The official Anthropic skills repo provides the core, while community additions like the **Markdown to EPUB Converter** and **Master Claude for Legal** pack extend into niche domains — the latter bundles NDA triage, multi-party diff, citation verification, and a weekly-status pattern for legal teams.


#### Development & Code Tools

The largest single category. Highlights include **Playwright Browser Automation** for automated web testing, **FFUF Web Fuzzing** for security scanning, **D3.js Visualization** for interactive charts, and **iOS Simulator** interaction. The **lean-ctx** skill manages context caching and AST-aware compression across Claude Code, Cursor, and Copilot. For full-stack projects, **great_cto** orchestrates 7 specialized subagents (tech-lead, senior-dev, QA-engineer, security-officer, devops, L3-support, project-auditor) covering the entire SDLC with 11 auto-detected project archetypes and 13 compliance frameworks.

The **Subagent-Driven Development** skill dispatches independent workers with code-review checkpoints between iterations — a pattern also built into OpenClaw and Hermes Agent. Others like **test-driven-development**, **using-git-worktrees**, and **Changelog Generator** solve specific pain points in the daily coding loop.


#### Data & Analysis

Skills for CSV analysis with auto-generated visualizations, read-only PostgreSQL querying with security guardrails, multi-step deep research via Gemini Deep Research, and recursive PhD-level research with disk checkpointing to survive context compaction.


#### Security & Systems

A growing subsection includes **Threat Hunting with Sigma Rules**, **computer forensics** analysis, secure file deletion, and metadata extraction. These transform an AI agent into a capable security assistant — querying detection rules, analyzing disk images, and sanitizing data — all within the same chat context.


#### CRM, Project Management & Enterprise Apps via Composio

Composio's own contribution is the largest chunk: pre-built workflow skills for **78 SaaS apps** — from Salesforce, HubSpot, and Zoho CRM to Jira, Linear, Notion, Slack, GitHub, Stripe, Shopify, and Zoom. Each skill comes with tool sequences, parameter guidance, known pitfalls, and real tool slugs discovered from Composio's API. These turn an AI agent into an automation hub that can act across the entire SaaS stack with no manual integration work.

The full list spans CRM & Sales, Project Management, Communication, Email, Code & DevOps, Storage & Files, Spreadsheets & Databases, Calendar & Scheduling, Social Media, Marketing, Support & Helpdesk, E-commerce, Design, Analytics, HR, and more.


### Cross-Platform — Skills Work Everywhere — Not Just Claude

The SKILL.md format has been adopted far beyond Anthropic's own ecosystem. OpenAI Codex, Cursor, Gemini CLI, Antigravity, Windsurf, and Claude Code all support the standard. This means a skill written once can be dropped into **any** compatible agent — your prompt engineering, workflow design, and guardrails travel with you between platforms.


### Getting Started — How to Use These Skills

On **Claude.ai**: click the skills icon (🧩) in the chat interface, then browse or upload. On **Claude Code**: place the skill folder in `~/.config/claude-code/skills/`. The skill activates automatically when its description matches your task. The repository includes detailed per-skill SKILL.md files with prerequisites, examples, and error-handling guidance — everything you need to pick up and run.

> **🧩** Skills are not the same as MCP servers. MCP handles auth and transport to external systems; skills handle behavior. In production you'll likely use both: MCP for connections, tools for actions, skills for workflows.


### Bottom Line — A Community-Driven Standard in the Making

The Awesome Claude Skills repo is more than a directory — it's a snapshot of where the AI agent ecosystem is heading. As agents become capable of more complex, multi-step work, the need for shareable, reusable *behavior packages* grows. The skill format competes directly with closed ecosystems like OpenAI's GPTs, but with the advantage of being open, file-based, and platform-agnostic.

With 1,000+ skills already published and contributions arriving daily, the library is well on its way to becoming the npm of AI agent workflows. Developers and power users can [browse the full collection on GitHub](https://github.com/ComposioHQ/awesome-claude-skills) — or submit their own.


## Primary sources

- [Awesome Claude Skills on GitHub](https://github.com/ComposioHQ/awesome-claude-skills)
- [Anthropic Skills Overview](https://www.anthropic.com/news/skills)

---
_Published by iyu (https://iyu.app) — the day's AI news, checked against primary sources and rewritten in plain language. Free to quote with attribution and a link to the canonical URL._