Custom Tools

Create and manage custom tools in MIRIX to extend your agent's capabilities with your own Python functions. They will be run in a sandbox environment.

Basic Tool Creation

Simple Function Tool

from mirix import Mirix

# Initialize memory agent
memory_agent = Mirix(api_key="your-google-api-key")

# Create a simple calculation tool
result = memory_agent.insert_tool(
    name="calculate_sum",
    source_code="def calculate_sum(a: int, b: int) -> int:\n    return a + b",
    description="Calculate the sum of two numbers",
    args_info={"a": "First number", "b": "Second number"},
    returns_info="The sum of a and b",
    tags=["math", "utility"] # "tags" is optional and not important
)

API Integration Tool

# Create a weather API tool
weather_tool_code = """def get_weather(city: str, api_key: str) -> str:
    import requests
    import json

    url = f"http://api.openweathermap.org/data/2.5/weather"
    params = {
        'q': city,
        'appid': api_key,
        'units': 'metric'
    }

    try:
        response = requests.get(url, params=params)
        response.raise_for_status()
        data = response.json()

        temp = data['main']['temp']
        description = data['weather'][0]['description']
        humidity = data['main']['humidity']

        return f"Weather in {city}: {temp}°C, {description}, humidity: {humidity}%"
    except Exception as e:
        return f"Error getting weather data: {str(e)}"
"""

result = memory_agent.insert_tool(
    name="get_weather",
    source_code=weather_tool_code,
    description="Get current weather information for a specified city",
    args_info={
        "city": "Name of the city to get weather for",
        "api_key": "OpenWeatherMap API key"
    },
    returns_info="Weather information string with temperature, description, and humidity",
    tags=["weather", "api", "external"] # "tags" is optional and not important
)

Tool Management Parameters

Method Signature

def insert_tool(
    self, 
    name: str, 
    source_code: str, 
    description: str, 
    args_info: Optional[Dict[str, str]] = None, 
    returns_info: Optional[str] = None, 
    tags: Optional[List[str]] = None, 
    apply_to_agents: Union[List[str], str] = 'all'
) -> Dict[str, Any]

Parameter Details

Parameter	Type	Description	Example
name	str	Function name identifier	"calculate_sum"
source_code	str	Python function code (without docstring)	"def func(x): return x * 2"
description	str	What the tool does	"Calculate the sum of two numbers"
args_info	Dict[str, str]	Argument descriptions	{"a": "First number", "b": "Second number"}
returns_info	str	Return value description	"The sum of a and b"
tags	List[str]	Categorization tags	["math", "utility"]
apply_to_agents	Union[List[str], str]	Which agents get the tool	'all' or ["chat_agent", "semantic_memory_agent"]

Available Agent Names

When specifying apply_to_agents, you can use any of these agent names:

• reflexion_agent - Handles self-reflection and metacognitive processes
• core_memory_agent - Manages core persistent memories
• episodic_memory_agent - Handles episodic and experiential memories
• procedural_memory_agent - Manages procedural knowledge and skills
• semantic_memory_agent - Handles semantic knowledge and facts
• meta_memory_agent - Manages meta-information about memories
• resource_memory_agent - Handles resource and file associations
• knowledge_vault_agent - Manages structured knowledge storage
• chat_agent - Handles conversational interactions
• background_agent - Processes background tasks and operations