Skip to content

image

AgentHelm

Production-Ready Orchestration for AI Agents.

AgentHelm is a lightweight Python framework for building AI agents with a focus on production-readiness. Built on DSPy, it provides the essential orchestration layer to make your agents observable, reliable, and safe.

Why AgentHelm?

In the rapidly evolving world of AI agents, many frameworks focus on rapid prototyping. AgentHelm is different—it's built for production environments where you need:

  • 🔍 Full observability of every tool call
  • 🛡️ Human-in-the-loop approval for sensitive actions
  • 🔄 Automatic rollbacks when things go wrong
  • 📊 Cost tracking and token usage monitoring

Installation

pip install agenthelm

Quick Start

CLI

# Initialize configuration
agenthelm init

# Run a simple task
agenthelm run "What is the capital of France?"

# Interactive chat
agenthelm chat

# Generate a plan
agenthelm plan "Build a web scraper" -o plan.yaml

# Execute the plan
agenthelm execute plan.yaml

Python SDK

import dspy
from agenthelm import ToolAgent, tool

# Define a tool
@tool()
def get_weather(city: str) -> str:
    """Gets the current weather for a given city."""
    if city == "New York":
        return "It is 24°C and sunny in New York."
    return f"Weather data not available for {city}."

# Create agent
lm = dspy.LM("mistral/mistral-large-latest")
agent = ToolAgent(name="weather_bot", lm=lm, tools=[get_weather])

# Run task
result = agent.run("What's the weather in New York?")
print(result.answer)

Key Features

🔧 Tool Decorator

@tool()
def search(query: str) -> str:
    """Search the web."""
    return f"Results for: {query}"

@tool(requires_approval=True)
def send_email(to: str, body: str) -> dict:
    """Send an email (requires approval)."""
    return {"status": "sent"}

📊 Observability

# View execution traces
agenthelm traces list

# Filter by status
agenthelm traces filter --status failed

# Export to Markdown
agenthelm traces export -o report.md -f md

🔗 MCP Integration

# Connect to MCP servers
agenthelm mcp list-tools uvx mcp-server-time
agenthelm mcp run uvx mcp-server-time -t "What time is it?"

🔄 Plan-Driven Execution

# Generate a multi-step plan
agenthelm plan "Build a todo app" -o plan.yaml

# Review and execute
agenthelm execute plan.yaml

Next Steps

License

MIT License - See LICENSE for details.