Tool Calling

​

Pre Configured Tools

1. Rest API Template
2. Webscraper Template
3. Custom tool Template

​

Drafting your Tool Definition

1. Name: Name you tool
2. Description: Describe both what the function does, when it should be called, and what parameters are required to call the function
3. Parameter Section: Define the JSON schema that should be outputted by the model when the model decides to make a function call.
   • The required properties will always provided by the model.

​

Drafting your Python Script**

1. The arguments for your function should match up with the appropriate properties in your tool definition. These arguments should be listed in the requiered attribute under parameters
   • "required": ["url", "selector"]
2. Only a single root function is allow
   • Nested Functions are allowed
3. Outputs from previous nodes or variables can be passed in using standard variable naming convention
   • <?REPLACE_WITH_REF_NODE_ID?>
4. Allowed functionality within Python Script:
   • Built-In Python Functions
     • int, str, isinstance, range, len, type, sum, round, ord, float, abs, min, max, all, any, sorted, enumerate, zip, filter, map, reversed, chr, divmod, pow, bin, hex, and oct
   • Supported Packages
     • json, requests, datetime, time, re, hashlib, bs4, typing, contextlib, decimal
     • NOTE: in order to use these packages, you will need to import them into your code i.e.

       • Code Example

         Copy

         import requests
         from bs4 import BeautifulSoup
         
         def perform_web_scrape(url, selector, attribute=None):
             try:
                 response = requests.get(url)
                 soup = BeautifulSoup(response.text, 'html.parser')
                 element = soup.select_one(selector)
                 result = element.get(attribute) if attribute else element.text.strip()
                 return {
                     "status": "Success",
                     "data": result
                 }
         
             except Exception as e:
                 return {
                     "status": "Failure",
                     "error": str(e)
                 }
         
         

​

Sample Scenarios:

• Enabling assistants to fetch data: an AI assistant needs to fetch the latest customer data from an internal system when a user asks “what are my recent orders?” before it can generate the response to the user
• Enabling assistants to take actions: an AI assistant needs to schedule meetings based on user preferences and calendar availability.
• Enabling assistants to perform computation: a math tutor assistant needs to perform a math computation.
• Building rich workflows: a data extraction pipeline that fetches raw text, then converts it to structured data and saves it in a database.

Tool Definition:

{
    "name": "perform_web_scrape",
    "description": "Perform a web scraping operation",
    "parameters": {
        "type": "object",
        "properties": {
            "url": {"type": "string"},
            "selector": {"type": "object"},
            "attribute": {"type": "object"}
        },
        "required": ["url", "selector"]
    }
}


Companion Python Script:

import requests
from bs4 import BeautifulSoup

def perform_web_scrape(url, selector, attribute=None):
    try:
        response = requests.get(url)
        soup = BeautifulSoup(response.text, 'html.parser')
        element = soup.select_one(selector)
        result = element.get(attribute) if attribute else element.text.strip()
        return {
            "status": "Success",
            "data": result
        }

    except Exception as e:
        return {
            "status": "Failure",
            "error": str(e)
        }