Tool Execution

The lifecycle of a tool call.

How Tools Work

Tools are the "hands" of an agent. They allow the LLM to interact with the outside world. In Bothive, every tool is a strictly typed function defined in TypeScript or automatically generated from an Integration.

The Execution Loop

1. Intent Detection

The LLM decides it needs to call a tool (e.g., `google.search`) to fulfill a query.

2. Parameter Validation

Bothive runtime validates the arguments against the tool's schema (e.g., checking if `query` is a string).

3. Execution & Networking

The actual code runs (or HTTP request is sent).

4. Observation

The tool result is fed back to the LLM as a system message.