autourgos-llmagentkit 1.0.4

An advanced LLM agent framework supporting ReAct and Tool Calling paradigms.

Autourgos LLM Agent Kit

Autourgos LLM Agent Kit is a powerful, lightweight framework for building advanced AI agents. It supports both ReAct (Reasoning + Acting) and Tool Calling paradigms, allowing developers to create intelligent agents that can reason through complex problems and execute tools effectively.

Features

• ReAct Agent: Implements the reasoning-trace loop (Thought → Action → Observation) for complex problem-solving.
• Parallel ReAct Agent: A powerful agent that combines the reasoning of ReAct with the speed of parallel tool execution. It overcomes the limitations of sequential execution by running multiple independent tools at once.
• Tool Calling Agent: Optimized for sequential tool execution without explicit reasoning overhead.
• Autourgos Core: A robust tool management system with the @tool decorator for easy function-to-tool conversion.
• Memory Integration: Optional conversation memory support.
• Customizable Prompts: Easily modify agent behavior with custom prompt templates.
• Verbose Logging: Detailed colored terminal output for debugging and monitoring agent thought processes.
• Agent History Logging: Automatically export a detailed Markdown file for each task sequence capturing all thoughts, tool executions, and results (enable via agent_history=True).

Installation

Install via pip:

pip install autourgos-llmagentkit

Quick Start

1. Define Tools

Use the @tool decorator from autourgos.core to turn Python functions into tools.

from autourgos.core import tool

@tool
def calculate(expression: str) -> str:
    """Evaluates a mathematical expression."""
    return str(eval(expression))

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

2. Create a ReAct Agent

The ReAct agent reasons about the user's request and decides which tools to call.

from autourgos.llmagentkit import Create_ReAct_Agent

# Initialize your LLM client (must implement generate_response method)
# llm = MyLLMClient(...)

agent = Create_ReAct_Agent(llm=llm, verbose=True)
agent.add_tools(calculate, search)

response = agent.invoke("What is the calculation for 25 * 4 and search for its significance?")
print(response)

3. Create a Tool Calling Agent

For more direct tasks where explicit reasoning steps are less critical.

from autourgos.llmagentkit import Create_ToolCalling_Agent

agent = Create_ToolCalling_Agent(llm=llm, verbose=True)
agent.add_tools(calculate)

response = agent.invoke("Calculate 100 / 5")
print(response)

4. Create a Parallel ReAct Agent

This agent thinks like a ReAct agent but executes independent tools simultaneously, offering the best of both worlds: transparent reasoning and high speed.

from autourgos.llmagentkit import Create_ParallelReAct_Agent

agent = Create_ParallelReAct_Agent(llm=llm, verbose=True)
agent.add_tools(calculate, search)

# The agent will call calculate("25 * 4") and search("...") in parallel
response = agent.invoke("Calculate 25 * 4 and search for its significance.")
print(response)

Project Structure

• autourgos.llmagentkit: Core agent implementations (ReAct, Tool Calling, and Parallel ReAct).
• autourgos.core: Tool management and decorators.

License

This project is licensed under a proprietary license. See the LICENSE file for details. Not for resale or commercial use without permission.