Crate motosan_agent_tool 

motosan-agent-tool — the Tool trait, runtime tool execution context, and a registry, plus a catalogue of built-in tools.

In 0.4 this crate is wired to motosan-agent-primitives for the wire-format types (ToolCall, ToolResult, ToolAnnotations, ContentBlock). See tool for the layering / migration story.

Re-exports

pub use error::Error;

pub use error::Result;

pub use registry::ToolRegistry;

pub use tool::Tool;

pub use tool::ToolContext;

pub use tool::ToolDef;

pub use tool::ToolOutput;

Modules

error

registry

tool

The Tool trait, execution context, and runtime tool output type.

tools

Built-in tool implementations.

Structs

ToolAnnotations

Capability metadata a tool publishes about itself.

ToolCall

An assistant-issued request to invoke a named tool.

ToolResult

The reply pairing for a ToolCall.

ToolSchema

Canonical LLM-facing tool declaration shared across the workspace: the name, human-readable description, and JSON-Schema input_schema the model sees. Layer-specific fields live on the wrappers — motosan_agent_tool::ToolDef adds the host-side internal_name, and motosan_ai::Tool adds the provider cache flag — both via #[serde(flatten)] so the on-the-wire shape is {name, description, input_schema, ...}.

Enums

ContentBlock

One slice of multimodal content inside a Message.

Value

Represents any valid JSON value.