From aradotso-trending-skills-37
Teaches Claude repeatable workflows via curated skills. Includes Composio integration for connecting to 500+ external apps like GitHub, Slack, and Gmail.
How this skill is triggered — by the user, by Claude, or both
Slash command
/aradotso-trending-skills-37:awesome-claude-skillsThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
```markdown
---
name: awesome-claude-skills
description: Curated collection of Claude Skills for Claude.ai, Claude Code, and Claude API — plus Composio integration for connecting Claude to 500+ apps
triggers:
- add a claude skill to my project
- connect claude to external apps
- install a claude skill
- create a CLAUDE.md skill
- use composio with claude
- set up claude code plugin
- automate workflows with claude skills
- find skills for claude code
---
# Awesome Claude Skills
> Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection.
A curated collection of Claude Skills that extend Claude's capabilities across Claude.ai, Claude Code, and the Claude API. Skills teach Claude repeatable, standardized workflows — from document processing and code tools to connecting Claude to 500+ external apps via Composio.
---
## What Are Claude Skills?
Claude Skills are markdown files (typically `SKILL.md` or placed in `.claude/skills/`) that give Claude specialized knowledge and workflows. They are installed into Claude Code as plugins or referenced via `CLAUDE.md`.
Skills enable Claude to:
- Follow consistent, repeatable workflows
- Integrate with external services (GitHub, Slack, Gmail, Notion, etc.)
- Apply domain-specific expertise (security, architecture, testing)
- Automate multi-step tasks autonomously
---
## Installation
### Install a Skill into Claude Code
```bash
# From a local directory
claude --plugin-dir ./my-skill-plugin
# Install Composio connect-apps plugin (most popular)
git clone https://github.com/ComposioHQ/awesome-claude-skills.git
cd awesome-claude-skills
claude --plugin-dir ./connect-apps-plugin
Skills referenced in CLAUDE.md are automatically picked up by Claude Code:
# Create a skills directory in your project
mkdir -p .claude/skills
# Copy or write a skill file
cp path/to/SKILL.md .claude/skills/my-skill.md
Reference it in your CLAUDE.md:
## Skills
@.claude/skills/my-skill.md
The connect skill lets Claude take real actions — send emails, create GitHub issues, post Slack messages, update Notion databases.
# 1. Install the plugin
claude --plugin-dir ./connect-apps-plugin
# 2. Run the setup wizard inside Claude Code
/connect-apps:setup
# 3. Paste your API key when prompted
# Get a free key at: https://platform.composio.dev
Set your API key as an environment variable:
export COMPOSIO_API_KEY="your_api_key_here"
Once connected, ask Claude naturally:
Send an email to [email protected] summarizing today's standup notes
Create a GitHub issue in my-org/my-repo titled "Fix login bug" with steps to reproduce
Post a message to the #deployments Slack channel: "v2.3.1 deployed to production"
Add a new page to my Notion database with today's meeting notes
import os
from composio import ComposioToolSet, App
toolset = ComposioToolSet(api_key=os.environ["COMPOSIO_API_KEY"])
# Get tools for specific apps
gmail_tools = toolset.get_tools(apps=[App.GMAIL])
github_tools = toolset.get_tools(apps=[App.GITHUB])
slack_tools = toolset.get_tools(apps=[App.SLACK])
# Use with an LLM (e.g., Anthropic)
import anthropic
client = anthropic.Anthropic()
all_tools = toolset.get_tools(apps=[App.GMAIL, App.GITHUB, App.SLACK])
response = client.messages.create(
model="claude-opus-4-5",
max_tokens=1024,
tools=all_tools,
messages=[
{
"role": "user",
"content": "Create a GitHub issue for the bug we discussed and notify the team on Slack"
}
]
)
# Handle tool calls
result = toolset.handle_tool_calls(response)
print(result)
| Skill | What it Does |
|---|---|
docx | Create/edit/analyze Word docs with tracked changes |
pdf | Extract text, tables, metadata; merge & annotate PDFs |
pptx | Read, generate, and adjust PowerPoint slides |
xlsx | Spreadsheet manipulation: formulas, charts, transformations |
# Install official Anthropic document skills
git clone https://github.com/anthropics/skills.git
ls skills/skills/ # docx, pdf, pptx, xlsx
# Changelog generator — transforms git commits to user-facing changelogs
# Place SKILL.md in .claude/skills/changelog-generator.md
# Then ask Claude: "Generate a changelog from the last 10 commits"
# MCP Builder — scaffold MCP servers
# Ask: "Create an MCP server for the Stripe API"
# Webapp Testing with Playwright
# Ask: "Test the login flow on localhost:3000 and take screenshots"
# postgres skill — safe read-only queries
# Configure your connection string:
export POSTGRES_CONNECTION_STRING="postgresql://user:pass@localhost:5432/mydb"
# Then ask Claude: "Show me the top 10 customers by revenue this month"
---
name: my-skill-name
description: One-line description of what this skill does
triggers:
- phrase users might say to invoke this
- another natural trigger phrase
- do the thing this skill handles
---
# My Skill Name
## Overview
What this skill does and when to use it.
## Instructions
Step-by-step guidance for Claude to follow.
## Examples
Concrete examples of inputs and expected outputs.
Install the built-in skill creator skill, then ask:
Create a skill that [describes your workflow]
Claude will scaffold a complete SKILL.md with frontmatter, instructions, and examples.
# skill_creator.py — programmatically generate skill stubs
import anthropic
import yaml
def create_skill(name: str, description: str, workflow: str) -> str:
client = anthropic.Anthropic()
prompt = f"""Create a Claude Skill SKILL.md file for the following:
Name: {name}
Description: {description}
Workflow: {workflow}
Include YAML frontmatter with name, description, and 6 triggers.
Include sections: Overview, When to Use, Instructions, Examples, Configuration."""
response = client.messages.create(
model="claude-opus-4-5",
max_tokens=2048,
messages=[{"role": "user", "content": prompt}]
)
return response.content[0].text
skill_content = create_skill(
name="api-rate-limit-handler",
description="Automatically handles API rate limiting with exponential backoff",
workflow="Detect 429 responses, implement backoff, retry requests, log outcomes"
)
with open("SKILL.md", "w") as f:
f.write(skill_content)
your-project/
├── CLAUDE.md # References skills with @path syntax
├── .claude/
│ └── skills/
│ ├── testing.md # TDD workflow skill
│ ├── git-workflow.md # Git branching/PR skill
│ └── api-integration.md # API patterns skill
└── connect-apps-plugin/ # Composio plugin directory
├── plugin.json
└── commands/
└── setup.js
# Project: My Application
## Active Skills
@.claude/skills/testing.md
@.claude/skills/git-workflow.md
## Project Context
This is a Python FastAPI application with PostgreSQL.
# iOS Simulator testing
git clone https://github.com/conorluddy/ios-simulator-skill
# D3.js data visualizations
git clone https://github.com/chrisvoncsefalvay/claude-d3js-skill
# Playwright browser automation
git clone https://github.com/lackeyjb/playwright-skill
# AWS CDK best practices
git clone https://github.com/zxkane/aws-skills
# EPUB generation from markdown
git clone https://github.com/smerchek/claude-epub-skill
# Install any skill by adding SKILL.md to .claude/skills/
cp claude-d3js-skill/SKILL.md .claude/skills/d3-visualization.md
# In your skill file
## Sub-Agent Pattern
When the task involves parallel work:
1. Identify independent subtasks
2. Dispatch separate agents for each using the Task tool
3. Collect results and synthesize
4. Run a final review pass
# Trigger the TDD skill
"Implement user authentication using TDD"
# Claude will:
# 1. Write failing tests first
# 2. Implement minimal passing code
# 3. Refactor while keeping tests green
# 4. Document the implementation
# Using the git-worktrees skill
# Ask: "Set up a worktree for the feature/payment-integration branch"
# Claude will run:
git worktree add ../project-payment feature/payment-integration
cd ../project-payment
# Work in isolation without affecting main branch
# Verify Claude Code can see your skill
cat CLAUDE.md # Should have @path references
# Check skill file syntax
head -20 .claude/skills/my-skill.md # Should start with --- YAML frontmatter
# Restart Claude Code after adding new skills
exit
claude
# Verify API key is set
echo $COMPOSIO_API_KEY
# Re-run setup
/connect-apps:setup
# Check connected apps
/connect-apps:list
# Test a specific connection
/connect-apps:test gmail
# Make triggers natural and varied in your frontmatter
triggers:
- generate a changelog # imperative
- create release notes # synonym
- what changed in this release # question form
- summarize git commits # descriptive
- prepare a changelog from git # verbose form
- update the changelog file # action-oriented
# When direct fetch returns 403, use the reddit-fetch skill pattern
# It routes through Gemini CLI as a fallback
# Ask: "Fetch the content from [url] using the reddit-fetch approach"
# Fork the repository
git clone https://github.com/ComposioHQ/awesome-claude-skills.git
cd awesome-claude-skills
# Create your skill directory
mkdir skills/my-awesome-skill
cat > skills/my-awesome-skill/SKILL.md << 'EOF'
---
name: my-awesome-skill
description: What your skill does in one line
triggers:
- natural phrase 1
- natural phrase 2
- natural phrase 3
- natural phrase 4
- natural phrase 5
- natural phrase 6
---
# My Awesome Skill
## Overview
...
EOF
# Submit a pull request
git checkout -b add-my-awesome-skill
git add skills/my-awesome-skill/
git commit -m "Add my-awesome-skill: brief description"
git push origin add-my-awesome-skill
# Open PR at github.com/ComposioHQ/awesome-claude-skills
npx claudepluginhub aradotso/trending-skillsReference for building Agent Skills with SKILL.md, covering structure, frontmatter, progressive disclosure, testing, and distribution across Claude surfaces.
Guides creation, improvement, auditing, testing, and distribution of Claude Code skills using Anthropic's official methodology.
Guides creation of Claude Code skills and slash commands: SKILL.md structure, frontmatter fields, invocation controls, commands vs skills, and best practices.