Discover & Deploy
AI Agents
The world's premier directory for autonomous agents. Discover, compare, and deploy specialized AI tailored for complex workflows.
What is Agent Shelf?
Agent Shelf is an open registry where developers publish, discover, and download AI agent definitions as Markdown files. Each agent is a reusable set of instructions with YAML metadata that works across AI coding tools including Claude Code, Cursor, Windsurf, and GitHub Copilot.
How It Works
Authors define agents using Markdown with YAML frontmatter — specifying a name, description, version, and category. Agents are published via the web UI, a CLI publish skill, or the MCP server. Users browse by category, search by keyword, and download agents directly into their AI tools.
Agent Categories
Agents span coding, marketing, writing, data analysis, DevOps, design, security, education, productivity, and more. Each agent is versioned with SemVer, so teams can pin stable versions while authors iterate on new capabilities.
All Agents32 agents
Workflow Automator
Designs and builds automated workflows that connect APIs, process data, handle approvals, and orchestrate multi-step tasks. Eliminates repetitive manual work.
agentshelfv1.1.0UI/UX Designer
Designs user interfaces, creates component specifications, builds design systems, and provides UX feedback. Generates accessible, responsive layouts following Material Design and modern design principles.
agentshelfv1.1.0Test Generator
Generates comprehensive unit tests, integration tests, and end-to-end tests. Analyzes code to identify edge cases, writes test suites, and improves coverage across any framework.
agentshelfv1.1.0Technical Writer
Creates developer documentation, API guides, tutorials, READMEs, and architecture decision records. Turns complex systems into clear, well-structured documentation.
agentshelfv1.1.0Social Media Strategist
Plans and creates social media content strategies, writes platform-specific posts, analyzes engagement patterns, and builds content calendars for consistent brand presence.
agentshelfv1.0.0SEO Content Writer
Creates search-engine-optimized blog posts, landing pages, and articles with proper keyword integration, meta tags, and content structure for maximum organic reach.
agentshelfv1.1.0Security Auditor
Performs security audits on codebases, dependencies, and infrastructure. Identifies vulnerabilities, suggests fixes, and helps achieve compliance with OWASP, SOC 2, and GDPR standards.
agentshelfv1.1.0Research Analyst
Conducts deep research on any topic — market analysis, competitive intelligence, literature reviews, and trend analysis. Synthesizes findings into structured, cited reports.
agentshelfv1.1.0Prompt Engineer
Designs, tests, and optimizes prompts for LLMs. Builds system prompts, few-shot examples, and evaluation frameworks to get reliable, high-quality AI outputs.
agentshelfv1.0.0Project Planner
Breaks down projects into actionable plans with milestones, dependencies, and realistic timelines. Manages scope, identifies risks, and keeps complex initiatives on track.
agentshelfv1.0.0Personal Finance Advisor
Helps individuals build budgets, plan savings strategies, analyze spending patterns, and make informed decisions about debt, investments, and financial goals.
agentshelfv1.0.0Onboarding Guide
Helps new developers get productive in unfamiliar codebases. Explains architecture, traces data flows, maps dependencies, and answers questions about how things work and why.
agentshelfv1.0.0Frequently Asked Questions
What is an AI agent definition?
An AI agent definition is a Markdown file with YAML frontmatter that gives an AI tool a specific persona, expertise, and workflow. It includes metadata like name, description, version, and category, plus freeform instructions in the Markdown body. Agent definitions work across tools like Claude Code, Cursor, and Windsurf.
How do I publish an agent to Agent Shelf?
You can publish agents three ways: upload a Markdown file through the web interface, use the publish skill from your AI coding tool (Claude Code, Cursor, Windsurf, etc.), or push via the MCP server. All methods require a GitHub account for authentication. Each publish creates a versioned, immutable snapshot.
Is Agent Shelf free to use?
Yes, Agent Shelf is completely free. You can browse, search, download, and publish agents at no cost. All you need is a GitHub account to publish or interact with agents (likes, comments). The registry is open and community-driven.
What AI tools work with Agent Shelf?
Agent Shelf agents work with any tool that supports the Agent Skills specification or the Model Context Protocol (MCP). This includes Claude Code, Cursor, Windsurf, GitHub Copilot, and 40+ other AI coding tools. Agents are plain Markdown files, so they can also be used manually with any AI assistant.