Open Agent Builder

Stale
GitHub TypeScript No License

Description

Visual workflow builder for AI agents powered by Firecrawl - drag-and-drop web scraping pipelines with real-time execution. Build agent workflows without coding.

Key Features

  • Visual drag-and-drop workflow builder with 8 core node types including Start, Agent, MCP Tools, Transform, If/Else, While Loop, User Approval, and End
  • Native Firecrawl integration for web scraping, searching, and converting websites to LLM-ready data
  • Real-time streaming execution with live workflow progress visualization
  • MCP protocol support for extensible external tool integration
  • Enterprise features: LangGraph execution engine, Clerk authentication, Convex database, and E2B sandboxed code execution
  • Template library with pre-built workflows for quick prototyping

Use Cases

💡 Building web scraping and data extraction pipelines without writing code
💡 Creating multi-step AI agent workflows with conditional branching and loops
💡 Automating research and content generation with human-in-the-loop approvals
💡 Developing data transformation and analysis pipelines powered by LLMs
💡 Rapid prototyping of AI agent applications using visual node-based programming

Tags

agent-builder visual-workflow web-scraping no-code typescript

Categories

⚡ Agent Tools 🔄 Workflow

Quick Start

1. Clone and install: `git clone https://github.com/firecrawl/open-agent-builder && cd open-agent-builder && npm install`
2. Set up Convex: `npx convex dev` (creates project and .env.local)
3. Configure Clerk auth and Firecrawl API key in .env.local
4. Run: `npm run dev:all` then visit http://localhost:3000
5. Sign up, add LLM API key in Settings → API Keys, click New Workflow, and try the Simple Web Scraper template
Visit GitHub

Related Projects