AI Tool Client

ibm_watsonx_gov.tools.clients.ai_tool_client.delete_tool(tool_id: str, inventory_id: str, **kwargs)

Delete the AI tool based on the tool_id and inventory_id

Parameters:

  • tool_id (str) – The ID of the tool to be deleted.

  • inventory_id (str) – The ID of the inventory to which the tool contains.

Returns:

AI Tool Deleted Successfully

Example:

from ibm_watsonx_gov.tools.clients.ai_tool_client import delete_tool
tool_id = '23dd7ec8-3d13-4f8f-a485-10278287bxxx'
inventory_id = '81b1b891-0ded-46ca-bc3f-155e98a15xxx'
response = delete_tool(tool_id=tool_id, inventory_id=inventory_id)
ibm_watsonx_gov.tools.clients.ai_tool_client.delete_tool_with_name(tool_name)

Method to delete the tool with tool_name

Parameters:

tool_name (str) – Name of the tool

ibm_watsonx_gov.tools.clients.ai_tool_client.get_tool(tool_id: str, inventory_id: str, **kwargs)

Retrieves the details of a specific tool based on the given tool ID and inventory ID.

Parameters:

  • tool_id (str) – The ID of the tool to be retrieved.

  • inventory_id (str) – The ID of the inventory to fetch the tool from.

Returns:

The processed result containing tool details.

Return type:

dict

Example:

from ibm_watsonx_gov.tools.clients.ai_tool_client import get_tool
tool_id = '23dd7ec8-3d13-4f8f-a485-10278287bxxx'
inventory_id = '81b1b891-0ded-46ca-bc3f-155e98a15xxx'
response = get_tool(tool_id=tool_id, inventory_id=inventory_id)
ibm_watsonx_gov.tools.clients.ai_tool_client.get_tool_info(tool_name: str, inventory_id: str = None) → dict

Method to retrieve tool information based on the tool name within the current inventory ID.

Parameters:

  • tool_name (str) – Name of the tool

  • inventory_id (str, optional) – Inventory id

Returns:

Returns the tool details based on the tool name.

Return type:

dict

Example:

from ibm_watsonx_gov.tools.clients.ai_tool_client import get_tool_info
tool_name = 'Google_search'
inventory_id = '81b1b891-0ded-46ca-bc3f-155e98a15xxx'
response = get_tool_info(tool_name=tool_name,inventory_id=inventory_id)
ibm_watsonx_gov.tools.clients.ai_tool_client.list_tools(service_provider_type: list[str] | str = None, category: list[str] | str = None, inventory_id: list[str] | str = None, framework: list[str] | str = None, tool_name: list[str] | str = None, run_time_details: list[str] | str = None, search_text: list[str] | str = None)

Retrieves a list of registered AI tools based on the provided filter criteria.

All parameters are optional. If no filters are provided, all tools will be listed.

Parameters:

  • service_provider_type (Union[list[str], str], optional) – Filter tools by one or more service provider types.

  • category (Union[list[str], str], optional) – Filter tools by one or more categories.

  • inventory_id (Union[list[str], str], optional) – Filter tools by one or more inventory IDs.

  • framework (Union[list[str], str], optional) – Filter tools by one or more frameworks (e.g., “langchain”).

  • tool_name (Union[list[str], str], optional) – Filter tools by one or more tool names.

  • run_time_details (Union[list[str], str], optional) – Filter tools by specific runtime details.

  • search_text – Filters tools based on matching keywords.

ibm_watsonx_gov.tools.clients.ai_tool_client.register_tool(payload: dict | ToolRegistrationPayload)

Registers a new tool by converting the provided payload to an ToolRegistrationPayload instance. And then call the tool registration endpoint.

Parameters:

payload (Union[dict, ToolRegistrationPayload]) – The payload to register the tool. It can either be a dictionary or an instance of ToolRegistrationPayload.

Returns:

The processed result from the tool registration API.

Return type:

dict

Examples:

  1. Using ToolRegistrationPayload model:
    from ibm_watsonx_gov.tools.clients import register_tool, ToolRegistrationPayload
post_payload = {
    "tool_name": "add",
    "description":"Use GooglSearch and return results for given query"
    "code": {
        "source_code_base64": '''
                def add(num1: int, num2: int):
                    return num1 + num2
        ''',
        "run_time_details": {
            "engine": "python 3.11.0"
        }
    },
    "schema":{
            "properties": {
                "num1": {
                    "description": "First number",
                    "title": "num1",
                    "type": "integer",
                },
                "query": {
                    "description": "Second number",
                    "title": "num2",
                    "type": "integer",
                }
            }
        }
 }
response = register_tool(ToolRegistrationPayload(**post_payload))
  2. Using Dict:
    from ibm_watsonx_gov.tools.clients.ai_tool_client import register_tool, Framework
post_payload = {
 "tool_name": "add",
 "code": {
    "source_code_base64": '''
            def add(num1: int, num2: int):
                return num1 + num2
    ''',
    "run_time_details": {
        "engine": "python 3.11.0"
    }
 },
 "schema":{
        "properties": {
            "num1": {
                "description": "First number",
                "title": "num1",
                "type": "integer",
            },
            "query": {
                "description": "Second number",
                "title": "num2",
                "type": "integer",
            }
        }
    }
 }
response = register_tool(payload=post_payload)
ibm_watsonx_gov.tools.clients.ai_tool_client.update_tool(tool_id: str, inventory_id: str, payloads: list[dict] | ToolUpdatePayload)

Patches the tool information based on the provided tool ID, inventory ID, and payloads.

Parameters:

  • tool_id (str) – The ID of the tool to be patched.

  • inventory_id (str) – The ID of the inventory to which the tool contains.

  • payloads (list[Union[dict, ToolUpdatePayload]]) – A list of dictionaries or ToolUpdatePayload instances containing the update information for the tool.

Returns:

The processed result from the tool update API.

Return type:

dict

Notes

  • Each item in the payloads list should be either a dictionary or an instance of ToolUpdatePayload.

  • If a dictionary is provided, it is converted into an ToolUpdatePayload instance.

Examples:

Using ToolUpdatePayload model:
from ibm_watsonx_gov.tools.clients.ai_tool_client import update_tool, ToolUpdatePayload, OPERATION_TYPE
patch_payload = [
 {
    "op": OPERATION_TYPE.REPLACE.value,
    "path": "/reusable",
    "value": False
},
{
    "op": OPERATION_TYPE.REPLACE.value,
    "path": "/metrics",
    "value": {
        "roc": "64"
    }
}
]
update_payload = ToolUpdatePayload(**patch_payload)
response = update_tool(payload = update_payload)
Using Dict:
from ibm_watsonx_gov.tools.clients.ai_tool_client import update_tool, OPERATION_TYPE
patch_payload = [
 {
    "op": OPERATION_TYPE.REPLACE.value,
    "path": "/reusable",
    "value": False
},
{
    "op": OPERATION_TYPE.REPLACE.value,
    "path": "/metrics",
    "value": {
        "roc": "64"
    }
}
]
response = update_tool(payload = patch_payload)