Note: This project is actually a fork of Open Agent Builder released by Firecrawl team. Some features are still very new and we welcome contributions and PRs!

The application is a visual workflow builder for creating AI agent pipelines powered by Composio's 10,000 + tools integration making a skill layer for your AI agents. You essentially build complex agent workflows with a drag-and-drop interface, then execute them with real-time streaming updates.

How It Works:

Drag-and-drop interface for building agent workflows
Real-time execution with streaming updates
8 core node types: Start, Agent, Tools, Transform, If/Else, While Loop, User Approval, End
MCP protocol support for extensible tool integration using Composio

We use:

Technology	Purpose
Composio	10,000 + tools integration making a skill layer for your AI agents
Next.js 16 (canary)	React framework with App Router for frontend and API routes
TypeScript	Type-safe development across the stack
LangGraph	Workflow orchestration engine with state management, conditional routing, and human-in-the-loop support
Convex	Real-time database with automatic reactivity for workflows, executions, and user data
Clerk	Authentication and user management with JWT integration
Tailwind CSS	Utility-first CSS framework for responsive UI
React Flow	Visual workflow builder canvas with drag-and-drop nodes
Anthropic	Claude AI integration with native MCP support (Claude Haiku 4.5 & Sonnet 4.5)
OpenAI	gpt-5 integration
Groq	Fast inference for open models
E2B	Sandboxed code execution for secure transform nodes

Installation & Setup

1. Clone the Repository

git clone https://github.com/patchy631/ai-engineering-hub.git
cd ai-engineering-hub/open-agent-builder
npm install

2. Set Up Convex (Database)

Convex handles all workflow and execution data persistence.

# Install Convex CLI globally
npm install -g convex

# Initialize Convex project
npx convex dev

This will:

Open your browser to create/link a Convex project
Generate a NEXT_PUBLIC_CONVEX_URL in your .env.local
Start the Convex development server

Keep the Convex dev server running in a separate terminal.

3. Set Up Clerk (Authentication)

Clerk provides secure user authentication and management.

Go to clerk.com and create a new application
In your Clerk dashboard:
- Go to API Keys
- Copy your keys
Go to JWT Templates → Convex:
- Click "Apply"
- Copy the issuer URL

Add to your .env.local:

# Clerk Authentication
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=pk_test_...
CLERK_SECRET_KEY=sk_test_...

# Clerk + Convex Integration
CLERK_JWT_ISSUER_DOMAIN=https://your-clerk-domain.clerk.accounts.dev

4. Configure Convex Authentication

Edit convex/auth.config.ts and update the domain:

export default {
  providers: [
    {
      domain: "https://your-clerk-domain.clerk.accounts.dev", // Your Clerk issuer URL
      applicationID: "convex",
    },
  ],
};

Then push the auth config to Convex:

npx convex dev

5. Optional: Configure Default LLM Provider

While users can add their own LLM API keys through the UI (Settings → API Keys), you can optionally set a default provider in .env.local:

# Anthropic Claude (Recommended - Native MCP support with Haiku 4.5 & Sonnet 4.5)
ANTHROPIC_API_KEY=sk-ant-...

# OpenAI GPT-5
OPENAI_API_KEY=sk-...

# Groq
GROQ_API_KEY=gsk_...

Important: For workflows using MCP tools, Anthropic Claude is currently the recommended provider as it has native MCP support.

6. Optional: E2B Code Interpreter

For advanced transform nodes with sandboxed code execution:

# E2B Code Interpreter (Optional)
E2B_API_KEY=e2b_...

Get your key at e2b.dev

Running the Application

# Terminal 1: Convex dev server
npx convex dev

# Terminal 2: Next.js dev server
npm run dev

Or run both with one command:

npm run dev:all

Visit http://localhost:3000

Get a FREE Data Science eBook 📖 with 150+ essential lessons in Data Science when you subscribe to our newsletter! Stay in the loop with the latest tutorials, insights, and exclusive resources. Subscribe now!

Contribution

Contributions are welcome! Feel free to fork this repository and submit pull requests with your improvements.

README.md

Open Agent Builder (with Composio)