🌐 OmniDaemon

Universal Event-Driven Runtime Engine for AI Agents

Run any AI agent. Any framework. One event-driven control plane.

Created by Abiola Adeshina • From the team behind OmniCore Agent

📚 Docs • 💡 Examples • 🏗️ Architecture • 🚀 Quick Start

---

🎯 What is OmniDaemon?

"Kubernetes for AI Agents" - A production-ready runtime that makes AI agents autonomous, observable, and fault-tolerant.

OmniDaemon runs each agent in its own isolated process (like containers), managed by an Agent Supervisor. This means your agents get production-grade reliability out of the box: process isolation, auto-restart, crash protection, and event-driven communication.

In 5 seconds:

• 🤖 Run any AI agent (OmniCoreAgent, Google ADK, LangChain, CrewAI, AutoGen, custom)
• 📨 Event-driven (agents react to events, not HTTP requests)
• 🔒 Process isolation (one agent crash won't kill others)
• 🔄 Auto-recovery (supervisors restart crashed agents)
• 🚀 Production-ready (retries, DLQ, metrics, scaling built-in)

💡 Why OmniDaemon exists: Most AI frameworks run everything in a single Python process. One crash kills your entire system. OmniDaemon solves this with process isolation and event-driven architecture. Read the full story →

---

🏭 Agent Supervisor: Production-Grade Process Isolation

The Agent Supervisor is OmniDaemon's core innovation. It manages each agent in an isolated process with automatic lifecycle management, error recovery, and health monitoring.

Benefits

Feature	What You Get
🔒 Fault Isolation	Agent A crashes? Agent B keeps running.
🔄 Auto-Recovery	Crashed agents restart automatically.
💾 Resource Safety	Clean memory/CPU boundaries per agent.
🎯 Process Per Agent	Like Kubernetes pods for AI agents.
❤️ Health Monitoring	Heartbeat checks, timeout handling.
📊 Observability	Metrics, logs, state tracking built-in.

Lifecycle State Machine

stateDiagram-v2
    [*] --> IDLE: Created
    IDLE --> STARTING: start()
    STARTING --> RUNNING: Process ready
    STARTING --> CRASHED: Start failed
    RUNNING --> STOPPING: stop()
    RUNNING --> CRASHED: Process died
    STOPPING --> STOPPED: Graceful shutdown
    CRASHED --> STARTING: Auto-restart
    CRASHED --> STOPPED: Max retries exceeded
    STOPPED --> [*]

Loading

📚 Deep dive: Agent Supervisor Architecture →

---

🚀 Quick Start

Get OmniDaemon running in 5 minutes with production-ready process isolation:

1. Install Redis (Event Bus)

# Using Docker (easiest)
docker run -d -p 6379:6379 --name redis redis:latest

# Verify
redis-cli ping  # Should return: PONG

2. Install OmniDaemon

# Using uv (recommended - 10-100x faster)
uv add omnidaemon

# Or using pip
pip install omnidaemon

3. Create Your Agent

# Create agent directory
mkdir my_first_agent
touch my_first_agent/__init__.py

Create agent (my_first_agent/agent.py):

def greeter_callback(message: dict):
    """Your agent runs here - in an isolated process!"""
    name = message.get("content", {}).get("name", "stranger")
    return {"reply": f"Hello, {name}! 👋"}

Create runner (agent_runner.py):

import asyncio
from omnidaemon import OmniDaemonSDK, AgentConfig
from omnidaemon.supervisor import create_supervisor_from_directory

sdk = OmniDaemonSDK()

async def main():
    # Create supervisor (runs agent in separate process)
    supervisor = await create_supervisor_from_directory(
        agent_name="greeter",
        agent_dir="./my_first_agent",
        callback_function="agent.greeter_callback"
    )
    
    await sdk.register_agent(
        agent_config=AgentConfig(topic="greet.user"),
        callback=supervisor.handle_event
    )
    
    await sdk.start()
    print("🎧 Agent running. Press Ctrl+C to stop.")
    
    try:
        while True:
            await asyncio.sleep(1)
    except KeyboardInterrupt:
        await sdk.shutdown()

if __name__ == "__main__":
    asyncio.run(main())

4. Run It!

python agent_runner.py

🎉 Your AI agent is now running in an isolated process with auto-restart!

📚 Next steps: Full Tutorial → | Complete Examples →

---

💡 Examples

See production-ready implementations in the examples/ directory:

Featured Examples

1. Multiple Agents with Supervisors - Run multiple supervised agents
2. Google ADK Integration - Full ADK agent with MCP tools
3. Content Moderation - Multi-agent pipeline
4. FastAPI Integration - HTTP + event-driven hybrid

Each example shows real-world patterns you'll use in production.

---

📚 Documentation

Core Concepts

• Why OmniDaemon? - The single process trap
• Event-Driven Architecture - Why EDA for AI agents
• Key Features - Deep dive into capabilities

Architecture

• Agent Supervisor - Process isolation & lifecycle
• System Overview - How everything fits together

How-To Guides

• Common Patterns - Production recipes
• Monitoring & Metrics - Observability
• Multi-Agent Systems - Coordination patterns

API Reference

• Python SDK - Complete SDK docs
• CLI Reference - Command-line tools

---

🎯 When to Use OmniDaemon

✅ Perfect For

• Background AI Agents - Autonomous agents reacting to events
• Event-Driven Workflows - Multi-step pipelines coordinated through events
• Distributed Multi-Agent Systems - Agents across different servers/runtimes
• Long-Running AI Tasks - Workloads that shouldn't block client requests
• Enterprise AI Ops - Production systems with retries, logs, monitoring

❌ Overkill For

• Simple HTTP APIs - Use FastAPI/Flask instead
• Pure Real-Time Chat Only - WebSockets/SSE alone are simpler
• Single-Shot Scripts - A basic Python script is sufficient

🆚 vs Alternatives

Tool	Use Case	vs OmniDaemon
Celery	Task queues	❌ Not AI-first, complex setup, no agent abstraction
AWS Lambda	Serverless functions	❌ Cold starts, time limits, vendor lock-in
Temporal	Workflow engine	❌ Heavy, complex, not AI-optimized
Airflow	DAG orchestration	❌ Batch-oriented, not real-time events
OmniDaemon	AI Agent Runtime	✅ AI-first, event-driven, any framework, production-ready

---

🔌 Pluggable Architecture

Switch backends by changing environment variables - zero code changes!

# Event Bus
EVENT_BUS_TYPE=redis_stream + REDIS_URL=redis://localhost:6379     # Current ✅
EVENT_BUS_TYPE=kafka + KAFKA_SERVERS=localhost:9092                 # Coming soon 🚧
EVENT_BUS_TYPE=rabbitmq + RABBITMQ_URL=amqp://localhost             # Coming soon 🚧

# Storage
STORAGE_BACKEND=redis + REDIS_URL=redis://localhost:6379            # Production ✅
STORAGE_BACKEND=json + JSON_STORAGE_DIR=./data                      # Development ✅
STORAGE_BACKEND=postgresql + POSTGRES_URL=postgresql://...          # Coming soon 🚧
STORAGE_BACKEND=mongodb + MONGODB_URI=mongodb://...                 # Coming soon 🚧

---

🌟 Resources

• 📖 Full Documentation - Complete guides and API reference
• 💡 Examples - Production-ready code samples
• 🏗️ Architecture - Deep dive into system design
• 🤝 Community - Get help and contribute
• 🐛 Issues - Report bugs or request features

---

📊 Features

Feature	Status
🔒 Process Isolation (Agent Supervisor)	✅ Production
📨 Event-Driven Architecture	✅ Production
🔄 Auto-Retry & DLQ	✅ Production
📊 Metrics & Monitoring	✅ Production
🎛️ CLI & HTTP API	✅ Production
🔌 Redis Event Bus	✅ Production
💾 Redis Storage	✅ Production
📁 JSON Storage	✅ Production
⚡ Kafka Event Bus	🚧 Coming Soon
🐰 RabbitMQ Event Bus	🚧 Coming Soon
🗄️ PostgreSQL Storage	🚧 Coming Soon
🍃 MongoDB Storage	🚧 Coming Soon

---

👨‍💻 About

Created by Abiola Adeshina and the OmniDaemon Team

From the creators of OmniCore Agent — building the future of event-driven AI systems

---

📄 License

MIT License - see LICENSE file for details

---

⭐ Star on GitHub • 🐛 Report Bug • 💡 Request Feature