bzzz/test/CHAT_INTEGRATION_README.md

Bzzz Chat-to-Code Integration Test Suite

This subproject demonstrates a complete chat-triggered workflow that integrates N8N with Bzzz agents for real-time code execution in ephemeral sandboxes.

🚀 Overview

The chat integration allows users to send natural language requests through a chat interface, which are then:

1. Parsed and validated by N8N workflow
2. Enhanced with LLM analysis for better task understanding
3. Sent to Bzzz agents for execution in isolated Docker sandboxes
4. Results returned to the chat with code artifacts and execution logs

📁 Files

test/
├── chat-to-code-integration.json   # N8N workflow definition
├── chat_api_handler.go              # Go API server for Bzzz integration
├── run_chat_api.sh                  # Build and run script
├── test_chat_api.py                 # Python test client
└── CHAT_INTEGRATION_README.md       # This file

🛠️ Setup Instructions

1. Prerequisites

Infrastructure:

• N8N instance running at https://n8n.home.deepblack.cloud/
• Ollama endpoints: WALNUT (192.168.1.27:11434), IRONWOOD (192.168.1.113:11434)
• Docker with Bzzz sandbox image: registry.home.deepblack.cloud/tony/bzzz-sandbox:latest

Dependencies:

# Go dependencies (auto-installed)
go get github.com/gorilla/mux

# Python dependencies
pip install requests

2. Start the Bzzz Chat API

# Build and run the API server
cd /home/tony/AI/projects/Bzzz
./test/run_chat_api.sh

This starts the API server on http://localhost:8080 with endpoints:

• POST /bzzz/api/execute-task - Execute task in sandbox
• GET /bzzz/api/health - Health check

3. Test the API

# Run the test suite
./test/test_chat_api.py

Expected output:

🧪 Bzzz Chat API Test Suite
========================================
🔍 Testing health check endpoint...
✅ Health check passed: {'status': 'healthy', 'service': 'bzzz-chat-api'}
🚀 Testing task execution...
✅ Task accepted: {'task_id': 9999, 'status': 'accepted'}
✅ All tests passed!

4. Import N8N Workflow

1. Open N8N Interface

   https://n8n.home.deepblack.cloud/
   

2. Import Workflow

   • Go to Workflows → Import from File
   • Upload chat-to-code-integration.json
   • Configure webhook URLs and API endpoints

3. Configure Endpoints

   {
     "bzzz_api_url": "http://localhost:8080/bzzz/api/execute-task",
     "ollama_walnut": "http://192.168.1.27:11434/api/generate",
     "ollama_ironwood": "http://192.168.1.113:11434/api/generate"
   }
   

4. Activate Workflow

   • Enable the chat trigger
   • Test with sample chat messages

💬 Usage Examples

Simple Requests

"Create a simple hello world function in Python"
"Build a calculator in JavaScript"
"Write a sorting algorithm in Go"

Structured Requests

Task: Implement user authentication API
Repo: https://github.com/myorg/api-server.git
Language: Python

Task: Fix memory leak in session handler
Repo: https://github.com/myorg/web-app.git
Language: JavaScript

Complex Requests

Build a React component that displays a todo list with add/remove functionality. 
Include proper TypeScript types and basic styling. Make it responsive and 
add unit tests using Jest.

🔄 Workflow Flow

1. Chat Input → User sends message via chat interface
2. Parse Request → Extract task details, repo, language
3. LLM Validation → Validate and enhance task description
4. Create Bzzz Request → Format for Bzzz API
5. Submit to Bzzz → Send to chat API handler
6. Sandbox Execution → Execute in isolated Docker container
7. Collect Results → Gather code artifacts and logs
8. Return to Chat → Send formatted results back

📊 API Reference

POST /bzzz/api/execute-task

Request:

{
  "method": "execute_task_in_sandbox",
  "task": {
    "task_id": 1001,
    "title": "Create hello world function",
    "description": "Create a simple Python function that prints hello world",
    "repository": {"owner": "test", "repository": "demo"},
    "git_url": "",
    "task_type": "development"
  },
  "execution_options": {
    "sandbox_image": "registry.home.deepblack.cloud/tony/bzzz-sandbox:latest",
    "timeout": "300s",
    "max_iterations": 5,
    "cleanup_on_complete": true
  },
  "callback": {
    "webhook_url": "https://n8n.home.deepblack.cloud/webhook/bzzz-result/1001",
    "include_artifacts": true
  }
}

Response:

{
  "task_id": 1001,
  "status": "accepted",
  "message": "Task accepted for execution"
}

Callback (Async):

{
  "task_id": 1001,
  "status": "success",
  "execution_time": "45.2s",
  "artifacts": {
    "files_created": [
      {
        "name": "hello.py",
        "size": 156,
        "content": "def hello_world():\n    print('Hello, World!')\n\nif __name__ == '__main__':\n    hello_world()",
        "language": "python"
      }
    ],
    "code_generated": "def hello_world():\n    print('Hello, World!')",
    "language": "python"
  },
  "execution_log": [
    {
      "step": 1,
      "action": "Starting task execution",
      "success": true,
      "timestamp": "2025-01-17T10:30:00Z"
    },
    {
      "step": 2,
      "action": "Created sandbox",
      "result": "Sandbox ID: abc123def456",
      "success": true,
      "timestamp": "2025-01-17T10:30:05Z"
    }
  ]
}

This chat integration provides a seamless bridge between natural language requests and actual code execution, making Bzzz agents accessible through familiar chat interfaces!