langchain-tools

To define custom LangChain tools, create callable functions with a name, description, and input schema. You can apply Python decorators to structure tool metadata, enabling models to invoke them based on context for data extraction or API interactions.

Yes, LangChain tools can be coroutines. You can define async tools to handle non-blocking tasks, allowing AI agents to execute operations like data fetching or code execution concurrently within Python workflows.

The best way to structure tool calling schemas is to provide clear input parameter definitions alongside a descriptive tool name. This structured metadata helps the model accurately determine when to call the tool and what arguments to provide.

Yes, some providers offer built-in server-side tools for LangChain agents, such as web search and code interpreter. These run on the provider's infrastructure and can be used alongside your custom Python tool definitions.

LangChain tools fail to trigger when their metadata is unclear. Models rely on accurate descriptions and deterministic usage patterns to decide when to call tools, so poorly defined schemas or missing return types prevent proper invocation.

No, Python decorators are not strictly required, but they provide efficient guidance on schemas and return types. Using decorators streamlines the process of binding structured tool definitions to models for AI agent workflows.