Build AI Agents that
matter.
A free, open source visual builder for AI Agent Skills, Prompts, and MCP Configurations. Pick an archetype, customize components, copy the code. Done in minutes.
Works with
Three steps, one file
From blank page to production-ready config
Pick an archetype
Agent Skill, MCP Config, System Prompt, OpenClaw Skill, or Copilot Instructions — choose the structure that fits your use case.
Choose a template
Start from a pre-made template, use the archetype defaults, or begin from scratch. Then add components to customize.
Edit, copy & ship
Customize components in the visual editor, preview the output live, then copy or download your file. Done.
New
Now with OpenClaw support
Build SKILL.md files for OpenClaw — the open source personal AI assistant with 200k+ stars. Create skills that work across WhatsApp, Telegram, Slack, Discord, and more.
5,700+
Community Skills
Published on ClawHub, the official OpenClaw skills registry.
15+ Channels
Multi-Platform
WhatsApp, Telegram, Slack, Discord, iMessage, Teams, and more.
Same Format
AgentSkills Compatible
SKILL.md with YAML frontmatter — works with VS Code Agent Skills too.
50+ Building Blocks
YAML frontmatter, tool definitions with JSON Schema, MCP server entries, environment variables, few-shot examples, and more.
Real Output Formats
Generates valid .agent.md, SKILL.md, .mdc, CLAUDE.md, .windsurfrules, and OpenClaw skill files — not pseudocode.
Instant Preview
See rendered markdown and raw source side-by-side as you build. Every edit updates live.
Open Source, No Lock-in
MIT licensed. No account, no telemetry, no server. Everything runs in your browser.
See it in action
Pick an archetype, customize components, preview your file in real-time
Frequently Asked Questions
Everything you need to know about building AI Agent files
An AI Agent Skill is a markdown file with YAML frontmatter that defines custom instructions, tools, and behaviors for AI assistants like GitHub Copilot. It allows you to extend the AI's capabilities with your own domain knowledge.
Model Context Protocol (MCP) allows AI models to securely connect to local or remote data sources. An MCP configuration is a JSON block that tells your AI client (like Claude Desktop or VS Code) how to connect to these servers.
Yes, completely free and open source under the MIT license. No account required, no hidden fees, no premium features locked behind a paywall.
Click Copy or Download in the builder, then place the file in the right location for your editor: .github/agents/ for VS Code, .cursor/rules/ for Cursor, CLAUDE.md for Claude Code, .windsurfrules for Windsurf, or skills/ for OpenClaw.
Archetypes define the structure and output format for a specific editor or use case — VS Code Agent Skills, Cursor Rules, Claude Code, OpenClaw Skills, MCP Server Configs, System Prompts, and more. Pick one, then choose from matching templates or start from scratch.
Absolutely. Every component is editable—change your agent's name, define custom JSON schemas for tools, add few-shot examples, or configure environment variables.
Your next agent starts here
No account needed. No data leaves your browser.
Free and open source — forever.
Open the Builder