Self-Healing Webserver

A dynamically evolving webserver that uses AI agents to handle unknown requests, automatically generates permanent handlers, and self-heals through continuous monitoring.

Architecture

• Gateway: Routes requests to handlers or AI interpreter
• AI Interpreter: Uses Ollama to understand and execute requests
• Coding Agent: Generates handler code from specifications
• Self-Healer: Monitors logs and generates corrections

Quick Start

Prerequisites

• Docker and Docker Compose
• At least 8GB RAM (for Ollama)

Start the Services

# Start all services
docker-compose up -d

# Pull the AI model (first time only)
./scripts/pull-model.sh llama3.2

# View logs
docker-compose logs -f

Test the Gateway

# Health check
curl http://localhost:3000/health

# Try an unhandled route (AI will interpret and respond)
curl http://localhost:3000/v1/users/1

# After AI handles it, a handler spec is generated
# The coding agent will process it and create a permanent handler

Services

Service	Port	Description
Gateway	3000	Main HTTP API
AI Interpreter	3010	Handles unknown requests
Coding Agent	-	Generates handlers
Self-Healer	-	Monitors and repairs
PostgreSQL	5432	Database
Ollama	11434	Local LLM

Development

# Start with live logs
docker-compose up

# Rebuild a specific service
docker-compose build gateway

# View service logs
docker-compose logs -f gateway

# Reset everything
./scripts/reset.sh

How It Works

1. Request arrives at Gateway
2. If handler exists → execute handler
3. If no handler → forward to AI Interpreter
4. AI Interpreter:
   • Introspects database schema
   • Uses Ollama to understand request intent
   • Executes SQL query
   • Returns response
   • Writes specification to queue
5. Coding Agent picks up specification:
   • Validates against current schema
   • Generates TypeScript handler
   • Updates registry
   • Triggers Gateway restart
6. Self-Healer monitors logs:
   • Detects errors and patterns
   • Generates correction specifications
   • Feeds back into Coding Agent

Directory Structure

selfhealserver/
├── docker-compose.yml
├── shared/                 # Shared types
├── services/
│   ├── gateway/           # HTTP Gateway
│   ├── ai-interpreter/    # AI request handling
│   ├── coding-agent/      # Code generation
│   └── self-healer/       # Error monitoring
├── scripts/               # Helper scripts
└── volumes/               # Local dev mounts (gitignored)

API Examples

# Get a user (AI will interpret first time)
curl http://localhost:3000/v1/users/1

# List products
curl http://localhost:3000/v1/products

# Create a user
curl -X POST http://localhost:3000/v1/users \
  -H "Content-Type: application/json" \
  -d '{"name": "John", "email": "john@example.com"}'

Configuration

Environment variables:

Variable	Default	Description
INTERNAL_SECRET	-	Secret for internal API calls
DATABASE_URL	-	PostgreSQL connection string
OLLAMA_URL	http://ollama:11434	Ollama API URL
LOG_LEVEL	info	Logging level

License

MIT "# self_gen_webserver"