Visual workflow builder for AI agents, Open Agent Builder enables users to design, test, and deploy complex agent pipelines using a drag-and-drop interface. It targets engineers and power users seeking to automate web scraping, data extraction, and multi-step AI processes without extensive coding. The project offers real-time execution and streaming updates, streamlining the development of sophisticated AI-driven workflows.

How It Works

Open Agent Builder leverages a React Flow-based visual canvas for workflow design, integrating a robust backend powered by Next.js, LangGraph for orchestration, and Convex for real-time data persistence. Core functionality includes a native Firecrawl integration for web scraping and data extraction, and support for the MCP protocol, allowing agents to interact with external tools. This architecture facilitates complex state management, conditional routing, and human-in-the-loop approvals within a unified, no-code environment.

Quick Start & Requirements

• Primary install/run command: npm install followed by npm run dev:all for development.
• Prerequisites: Node.js 18+, Firecrawl API key, Convex account, Clerk account. LLM API keys (Anthropic Claude recommended for MCP) and an E2B API key (optional) can be configured via UI or environment variables.
• Setup: Requires cloning the repository, installing dependencies, setting up Convex and Clerk accounts, and configuring environment variables (.env.local) with necessary API keys.
• Links: Documentation, Examples, Firecrawl (firecrawl.dev), Convex (convex.dev), Clerk (clerk.com), E2B (e2b.dev).

Highlighted Details

• Visual Workflow Builder with 8 core node types: Start, Agent, MCP Tools, Transform, If/Else, While Loop, User Approval, End.
• Native Firecrawl integration for web scraping and searching, powered by the MCP protocol.
• LangGraph execution engine for reliable state management, conditional routing, and human-in-the-loop support.
• Convex database for real-time workflow and execution data persistence.
• Clerk authentication for secure multi-user access.
• Programmatic execution via API endpoints with Server-Sent Events (SSE) streaming.

Maintenance & Community

The project is actively under development, with a roadmap indicating ongoing work on MCP support for additional LLM providers, OAuth authentication, and enhanced workflow features. Contributions and Pull Requests are welcomed. Links to the GitHub repository and Firecrawl are provided.

Licensing & Compatibility

This project is licensed under the MIT License, permitting commercial use and integration into closed-source projects.

Limitations & Caveats

The project is in active development, with some features still in progress. MCP support for OpenAI and Groq is coming soon, and complex loop patterns and custom node types are partially supported or planned for future releases. E2B Code Interpreter functionality is also noted as partial.