types

Structs

`ChatMessage`

Provider-neutral chat message supporting text, tool calls, and tool results.

Each provider maps this internal representation to its own wire format:

OpenAI: tool_calls in assistant messages, role: "tool" for results
Anthropic: tool_use content blocks, tool_result content blocks

Fields

Field	Type	Description
`role`	`String`
`content`	`Option<String>`
`tool_calls`	`Option<Vec<ToolCallRequest>>`	Tool calls requested by the assistant (present in assistant messages).
`tool_call_id`	`Option<String>`	ID of the tool call this message is responding to (present in tool-result messages).
`name`	`Option<String>`	Function name for tool-result messages.

`ToolCallRequest`

A tool call as requested by the LLM in an assistant message.

arguments is normalized to serde_json::Value internally — OpenAI sends arguments as a JSON string, Anthropic sends input as a JSON object. Each provider normalizes on ingest and serializes back to its wire format.

Fields

Field	Type	Description
`id`	`String`
`name`	`String`
`arguments`	`serde_json::Value`

`ToolSchema`

Fields

Field	Type	Description
`name`	`String`
`description`	`Option<String>`
`parameters`	`serde_json::Value`
`strict`	`Option<bool>`

`LlmRequest`

Fields

Field	Type	Description
`model`	`String`
`messages`	`Vec<ChatMessage>`
`tools`	`Option<Vec<ToolSchema>>`
`tool_choice`	`Option<ToolChoice>`
`temperature`	`Option<f32>`
`max_tokens`	`Option<u32>`
`extensions`	`Option<serde_json::Map<String, serde_json::Value>>`

`Usage`

Fields

Field	Type	Description
`prompt_tokens`	`Option<u32>`
`completion_tokens`	`Option<u32>`
`total_tokens`	`Option<u32>`

`LlmChoice`

Fields

Field	Type	Description
`index`	`u32`
`message`	`ChatMessage`
`finish_reason`	`Option<String>`

`ToolCall`

Fields

Field	Type	Description
`id`	`Option<String>`
`call_id`	`Option<String>`
`name`	`String`
`arguments`	`serde_json::Value`

`LlmResponse`

Fields

Field	Type	Description
`id`	`Option<String>`
`created`	`Option<u64>`
`model`	`Option<String>`
`choices`	`Vec<LlmChoice>`
`usage`	`Option<Usage>`
`tool_calls`	`Option<Vec<ToolCall>>`

Enums

`ToolChoice`

How the model should choose tools (OpenAI / Anthropic wire formats mapped internally).

Variants

Variant	Description
`Auto`
`None`
`Required`
`Function(String)`

Methods

`to_openai_value`

fn to_openai_value(&self) -> serde_json::Value

OpenAI Chat Completions / Responses tool_choice JSON.

`to_anthropic_value`

fn to_anthropic_value(&self) -> serde_json::Value

Anthropic Messages tool_choice JSON.