microAgents 0.1.0

A lightweight framework for building AI agents with XML-style tool calls

microAgents Framework

A lightweight framework for building AI agents with XML-style tool calls.

Features

• XML-style tool calling format for clear parameter structure
• Support for OpenAI-compatible LLM APIs
• Simple message history management
• Type-safe tool parameter handling
• Easy-to-use agent configuration

Installation

pip install microAgents

Quick Start

from microAgents.core import MicroAgent, Tool, MessageStore
from microAgents.llm import LLM

# Define a tool
def add_numbers(a: float, b: float) -> float:
    return a + b

# Create an agent
llm = LLM(base_url="https://api.openai.com/v1")
agent = MicroAgent(
    llm=llm,
    prompt="You are a math assistant",
    toolsList=[Tool("add", "Add two numbers", add_numbers)]
)

# Use the agent
message_store = MessageStore()
response = agent.execute_agent("What is 5 plus 3?", message_store)
print(response)

Tool Call Format

Tool use is formatted using XML-style tags. The tool name is enclosed in opening and closing tags, and each parameter is similarly enclosed within its own set of tags:

<TOOL_CALLS_NEEDED>
<tool_name>
<param1>value1</param1>
<param2>value2</param2>
</tool_name>
</TOOL_CALLS_NEEDED>

Multiple tool calls can be included:

<TOOL_CALLS_NEEDED>
<add>
<a>5</a>
<b>3</b>
</add>
<multiply>
<a>4</a>
<b>2</b>
</multiply>
</TOOL_CALLS_NEEDED>

Examples

Check out the examples/ directory for complete examples:

• math_demo.py: Basic math operations using tool calls

License

MIT License