skills  by anthropics

Dynamic instruction sets for enhancing AI task performance

Created 1 month ago
15,335 stars

Top 3.2% on SourcePulse

GitHubView on GitHub
Project Summary

This repository provides a collection of example "Skills" designed to dynamically enhance Anthropic's Claude AI's capabilities for specialized tasks. Targeted at developers and power users, these modular, instruction-based extensions allow Claude to perform repeatable actions, ranging from creative content generation and data analysis to enterprise workflows and complex document manipulation, thereby improving its performance and utility in specific domains.

How It Works

Skills are implemented as self-contained folders, each containing a SKILL.md file. This file houses instructions, metadata, and resources that Claude dynamically loads to execute specific tasks. This approach allows for task specialization without altering Claude's core model, enabling repeatable, context-aware actions and providing a flexible framework for extending AI functionality.

Quick Start & Requirements

  • Usage:
    • Claude Code: Register via /plugin marketplace add anthropics/skills.
    • Claude.ai: Available to paid plan users.
    • Claude API: Integrate through the Skills API Quickstart.
  • Prerequisites: Access to Claude Code, Claude.ai (paid tier), or the Claude API. No specific hardware or non-default software dependencies are listed for using the example skills.
  • Development: Use template-skill as a starting point for creating custom skills, requiring a SKILL.md with YAML frontmatter (name, description) and markdown instructions.

Highlighted Details

  • Diverse Examples: Includes skills for algorithmic art, Slack GIF creation, web app testing, artifact building (React/Tailwind), MCP server generation, brand guideline application, internal communications, and theme styling.
  • Document Skills: The document-skills/ subdirectory offers advanced examples for creating and manipulating DOCX, PDF, PPTX, and XLSX files, demonstrating complex file format and binary data handling.
  • Skill Development: Provides skill-creator guidance and a template-skill for building new, custom extensions.

Maintenance & Community

The repository is maintained by Anthropic, as indicated by the inclusion of official brand skills. While the repository is public, specific community channels (like Discord/Slack), active contribution guidelines, or a public roadmap are not detailed in the provided README.

Licensing & Compatibility

  • Example Skills: Licensed under Apache 2.0, permitting commercial use and modification.
  • Document Skills: Labeled as "source-available, not open source," suggesting potential usage restrictions beyond standard open-source licenses, though these are not specified.
  • Compatibility: Apache 2.0 licensed skills are generally compatible with commercial applications. The "source-available" nature of document skills requires careful review for integration into proprietary systems.

Limitations & Caveats

These skills are presented as reference examples for inspiration and education, not as production-ready solutions for organization-specific workflows. Claude's actual implementation and behavior may differ from the examples shown. The document skills are explicitly noted as "point-in-time snapshots" and are not actively maintained. Thorough testing in a target environment is recommended before relying on any skill for critical tasks.

Health Check
Last Commit

2 weeks ago

Responsiveness

Inactive

Pull Requests (30d)
47
Issues (30d)
32
Star History
15,695 stars in the last 30 days

Explore Similar Projects

Starred by Zhiqiang Xie Zhiqiang Xie(Coauthor of SGLang), Eric Zhu Eric Zhu(Coauthor of AutoGen; Research Scientist at Microsoft Research), and
3 more.

Trace by microsoft

0.7%
669
AutoDiff-like tool for end-to-end AI agent training with general feedback
Created 1 year ago
Updated 2 months ago
Feedback? Help us improve.