Utility tools

​

Composio Converter Tools

This module provides tools for converting Composio Apps and Actions into Galadriel Tools.

​

convert_action(api_key: str, action: str) -> Tool

Converts a single Composio action into a Galadriel Tool.

​

Parameters:

• api_key (str): Composio API key for authentication.
• action (str): Name of the Composio action to convert.

​

Returns:

• Tool: The converted Galadriel Tool.

​

Example:

converted_tool = convert_action("your_api_key", "some_action")

​

convert_app(api_key: str, app: App) -> list[Tool]

Converts all tools from a Composio App into Galadriel Tools.

​

Parameters:

• api_key (str): Composio API key for authentication.
• app (App): The Composio App to convert.

​

Returns:

• list[Tool]: List of converted Galadriel Tools.

​

Example:

converted_tools = convert_app("your_api_key", some_app)

​

Retriever Tool

This module provides semantic search functionality for retrieving relevant documents or document segments based on user queries. It uses the BM25 algorithm for ranking document relevance.

​

RetrieverTool(docs: List[Document], **kwargs)

A tool for semantic document retrieval using BM25 ranking.

​

Parameters:

• docs (List[Document]): List of documents to index for retrieval. Documents should be pre-processed and split if needed.
• **kwargs: Additional arguments passed to the parent Tool class.

​

Attributes:

• name (str): Tool identifier for the agent system (set to “retriever”).
• description (str): Description of the tool’s functionality.
• inputs (dict): Schema for the required input parameters.
• output_type (str): Type of data returned by the tool (set to “string”).
• retriever (BM25Retriever): The underlying BM25 retrieval engine.

​

Methods:

​

forward(query: str) -> str

Performs a semantic search across the document collection.

Parameters:

• query (str): The search query string. This should be semantically close to your target documents. Use the affirmative form rather than a question.

Returns:

• str: A formatted string containing the retrieved documents, with each document prefixed by its index.

Raises:

• AssertionError: If the query is not a string.

​

Example:

retriever_tool = RetrieverTool(docs=some_documents)
result = retriever_tool.forward("Find information about AI agents")