Key Capabilities
Everything you need to build sophisticated multi-agent applications
🏗️
Hierarchical Agent Management
Create manager and worker agents with unlimited nesting. Build complex multi-agent systems with clear hierarchies and responsibilities. Share workers across multiple parents for maximum efficiency.
🚀
Enterprise-Ready
Production-grade with built-in error handling, logging, and monitoring. Configurable timeouts, retries, and isolation ensure reliability at scale with millions of users per day.
🔧
Tool Integration
Easy tool creation with OpenAI-compatible schemas. Register Python functions as tools with automatic validation. Support for async operations and complex tool chains.
🔒
Complete Isolation
Full isolation between agent instances ensures security and stability. Worker agents operate independently with controlled message passing and no internal state leakage.
🤝
OpenAI Compatible
Works seamlessly with OpenAI message formats and responses. Drop-in replacement for existing OpenAI integrations with enhanced multi-agent capabilities.
📊
Visual Tracing
Zero-overhead tracing with beautiful execution visualizations. Generate Mermaid diagrams, interactive HTML dashboards, and export execution data for analysis.
Architecture Overview
Hierarchical multi-agent system with unlimited nesting and tool integration
Manager Agent
Worker Agent
Tool
Agent Communication
Tool Registration
Workflow Lifecycle
Intelligent orchestration with Planning Agent and Tool/Agent Calls
Quick Start
Get up and running with AgentRouter in minutes
1
Install
Install AgentRouter SDK from PyPI with a simple pip command
2
Configure
Set up your manager agent with API key and model preferences
3
Build
Create workers, register tools, and define your workflow
4
Execute
Run your multi-agent system and handle complex tasks
# Install AgentRouter
pip install agentrouter
# Create your first multi-agent system
from agentrouter import ManagerAgent, WorkerAgent, tool
# Initialize Manager Agent
manager = ManagerAgent(
name="customer_service_manager",
api_key="your-api-key",
model="usf-mini",
backstory="Senior customer service manager",
goal="Resolve customer issues efficiently"
)
# Create Worker Agent
tech_support = manager.create_worker(
name="Technical_Support",
role="Diagnoses hardware/software issues, provides troubleshooting steps, and escalates complex technical problems" # REQUIRED
)
# Define a Tool
@tool(
schema={
"type": "function",
"function": {
"name": "search_knowledge_base",
"description": "Searches technical documentation for solutions, returns troubleshooting steps and fixes",
"parameters": {
"type": "object",
"properties": {
"query": {"type": "string"}
},
"required": ["query"]
}
}
}
)
def search_knowledge_base(query: str):
return {"results": f"Found solutions for: {query}"}
# Register the tool
tech_support.register_tool(search_knowledge_base)
# Execute a task
messages = [
{"role": "user", "content": "My laptop screen is flickering"}
]
response = await manager.execute(messages)
print(response)
🎯
Production Ready
Enterprise-grade with built-in error handling and monitoring
🔄
Fully Configurable
Customize timeouts, retries, and execution parameters
📈
Scalable
Handle millions of requests with isolated agent instances