MatrixSwarm Master Class: Factory Injection

MatrixSwarm agents don’t hardcode their behavior. Instead, they inject it at runtime using factories.

This allows agents to be reused, reprogrammed, and rewired without rewriting the class — making logic deployment modular, like loading ammo.

🧩 What is a Factory?

A factory is a Python module with an attach(agent, config) function.

threading.Thread(target=reflex_loop, daemon=True, name="agent_reflex").start()

The agent loads this during boot using the resolve_factory_injections() hook built into BootAgent.

🔧 Declaring Factories

Factories are defined in the swarm tree:

"config": {
  "role": "alert_listener",
  "factories": {
    "alert.subscriber": {
      "levels": ["critical", "warning"],
      "chat_id": "...",
      "bot_token": "..."
    }
  }
}

This will load:

agent/your_agent/factory/alert/subscriber/__init__.py

When Matrix spawns the agent, it will automatically call:

import agent.your_agent.factory.alert.subscriber
subscriber.attach(agent, config["factories"]["alert.subscriber"])

📡 Why it Matters

• Hot-swappable logic with no restarts
• One agent class, infinite behaviors
• Perfect for relays, listeners, responders, even AI stacks

Factory injection makes MatrixSwarm modular. You don’t build new agents. You train them to behave differently — live.