Toolhouse - Store

Web search allows your LLM to perform web searches using a search engine. This tool returns the most up to date information about the search query your LLM generates, along with a summary of the page's content for each result. By default, Web search returns up to 5 results, but your LLM can specify a different number of results. To do so, you can prompt your LLM to pass a 'num_results' argument to your query.

PDF2CSV is a tool that extracts all tables from a PDF document and returns a json that represents an array of CSV files encoded in.

Random emoji gives the LLM a random emoji (and sometimes, a random Unicode character). You can use this tool to test out your Toolhouse integration without consuming execution credits.

This tool is the same as Pinecone Storage, except it allows you to connect directly to your Pinecone instance. Pinecone Vector Upsert enables your LLM to store information. When your LLM calls Pinecone Vector Upsert, it will trigger Toolhouse and it will create embeddings for you. This way, you can equip your LLM with the ability to store data like user data, context, or results from other tools. You can use Pinecone Vector Upsert to retrieve and reference data across sessions.