ufokechukwu/ignode_ai_agent_builder repository overview

⁠IGNode AI Agent Builder

A visual, low-code platform for designing AI agent workflows and generating production-ready code for LangGraph Python.

Demo: Building a Multi Agent HR-Support Ticketing System⁠

⁠Overview

IGNode AI Agent Builder bridges the gap between no-code workflow design and enterprise-grade AI agent deployment. Design sophisticated AI agent workflows on an interactive canvas, then generate production-ready Python code for your preferred framework.

Docker hosted visual workflow designer for building, testing, and deploying AI agents powered by LangGraph.

⁠Quick Start

⁠Option 1: Docker Compose (Recommended)

Includes PostgreSQL, MQTT broker, and IoT simulator out of the box.

git clone https://bitbucket.org/letscodewithfrancis/docker-ignode-ai-agent-builder.git
cd docker-ignode-ai-agent-builder
docker compose up -d

Open http://localhost:7789⁠ in your browser.

⁠Option 2: Docker Run (Standalone)

Run the agent builder with just Docker. You need to provide your own PostgreSQL database.

1. Start a PostgreSQL database (skip if you already have one):

docker run -d --name agent-builder-db \
  -e POSTGRES_DB=agent_builder \
  -e POSTGRES_USER=postgres \
  -e POSTGRES_PASSWORD=postgres \
  -p 5432:5432 \
  postgres:16-alpine

2. Start the agent builder:

docker run -d --name ignode-agent-builder \
  -p 7789:80 \
  -e AGENT_BUILDER_DB_HOST=host.docker.internal \
  -e AGENT_BUILDER_DB_PORT=5432 \
  -e AGENT_BUILDER_DB_NAME=agent_builder \
  -e AGENT_BUILDER_DB_USER=postgres \
  -e AGENT_BUILDER_DB_PASSWORD=postgres \
  -e DEPLOY_ENABLED=true \
  -e JUPYTER_ENABLED=true \
  -v agent_deployments:/app/agent-runner/deployments \
  -v agent_data:/app/agent-runner/data \
  -v rag_documents:/data/rag-documents \
  -v notebooks:/notebooks \
  ufokechukwu/ignode_ai_agent_builder:latest

Open http://localhost:7789⁠ in your browser.

⁠What's Included

⁠Architecture

Single all-in-one container (ignode-ai-agent-builder) running:

• React Frontend — drag-and-drop visual workflow designer
• .NET 8 Backend API — workflow CRUD, code generation, settings, templates
• Python Agent Runner — deploys and runs LangGraph agents (FastAPI/uvicorn)
• JupyterLab — interactive notebook for testing generated code

Plus supporting containers:

• PostgreSQL 16 — database for workflows, settings, tools, RAG documents
• Mosquitto MQTT — broker for IoT agent workflows
• IoT Simulator — publishes sample MQTT data for testing IoT templates

⁠Features

• 18 workflow templates (Getting Started, AI Routing, HITL, Security, IoT, RAG, MCP, and more)
• Drag-and-drop node editor with 16 node types
• LangGraph Python code generation and one-click deployment
• Built-in chat testing with SSE streaming
• Human-in-the-Loop (approval, input, selection, review modes)
• RAG knowledge base with document upload
• MCP server connections
• Custom Python function nodes
• Data Formatter with dot-notation variable mapping
• Security Gate node with rule-based filters + optional LLM verification
• Dark/Light mode

⁠Configuration

⁠LLM Provider Setup

The agent builder does not include LLM API keys. After launching:

1. Open the app at http://localhost:7789⁠
2. Create or open a workflow
3. Go to Menu > Settings > LLM Providers
4. Add your provider (OpenAI, Anthropic, Google, etc.) with your API key
5. Select the provider on each Agent node

⁠Environment Variables

⁠Custom Postgres Password

POSTGRES_PASSWORD=mysecretpassword docker compose up -d

⁠Versions

⁠Latest (v2.2.0)

• Host up to 3 deployed agents at once, each callable at /agent/{id}/invoke
• Agent Operations: concurrent chat sessions, type-ahead input, per-message local times, unified menu
• Custom (OpenAI-compatible) LLM providers; token/usage caps + resume accounting
• Image/vision input at the wire; full ctx SDK usable in the runtime and JupyterLab
• Templates + ctx SDK help chapters; 36 builtin templates

⁠v2.0.0

• All-in-one container with external PostgreSQL
• Single port access (7789)
• 18 workflow templates with auto-provisioning (tools, MCP, RAG)
• RAG storage mode (local/api)
• Gate node with RAG/tools in LLM verification
• Multi-platform: linux/amd64 + linux/arm64

⁠v0.1.0

To run the previous version:

docker compose -f docker-compose.v0.1.0.yml up -d

⁠Stopping

docker compose down

To also remove data volumes:

docker compose down -v

⁠Tutorial

Youtube: IGNode AI Agent Builder⁠

⁠Support

For issues and feature requests, please open an issue on GitHub.

⁠License

• Free for personal use.
• Commercial use requires a license.

Built with ❤️ for the AI Agent community