nFactorial

Build agents that run directly or on a distributed runtime.

nFactorial is an agent framework with a Redis-backed orchestrator for reliable multi-agent systems. It makes the following straightforward to implement:

• Direct and queued execution: Run agents locally with agent.run(...) or on a distributed Redis-backed runtime.
• Agent reliability: Automatic retries, backoff strategies, and recovery of dropped tasks from crashed workers.
• In-flight task management: Cancel, steer, wake, branch, and monitor running tasks.
• Subagents and waits: Spawn child tasks and pause on sleeps, activity, signals, jobs, or approval hooks.
• Runtime coordination: Exchange direct or group messages, read inboxes, and process receipts.
• Observability: Built-in metrics dashboard and comprehensive logging.

Installation

pip install nfactorial

Quick Start

import asyncio

from factorial import Agent, gpt_41


def get_weather(location: str) -> str:
    return f"The weather in {location} is sunny and 72°F"


agent = Agent(
    instructions="You help users get weather information.",
    model=gpt_41,
    tools=[get_weather],
)


async def main() -> None:
    result = await agent.run("What's the weather in San Francisco?")
    print(result.output)


asyncio.run(main())

When you need waits, hooks, subagents, messaging, retries, or distributed workers, register the agent with an Orchestrator and enqueue tasks instead.

Usage Examples

IDE Coding Agent

docs/docs/examples/code_agent.md

Multi-Agent

docs/docs/examples/multi_agent.md

Group Chat Agent

docs/docs/examples/group_chat_agent.md

Mafia Game Agent

examples/mafia_game_agent

Deliberations.ai

Check it out