Enum ContentBlock 

#[non_exhaustive]
pub enum ContentBlock {
    Text {
        text: String,
    },
    Image {
        source: ImageSource,
    },
    Reasoning {
        format: ReasoningFormat,
        text: String,
        signature: Option<String>,
        payload: Option<Value>,
    },
    ToolUse {
        id: String,
        name: String,
        input: Value,
    },
    ToolResult {
        tool_use_id: String,
        content: Vec<ResultChunk>,
        is_error: bool,
    },
}

A typed piece of message content, modeled on Anthropic’s content blocks.

#[non_exhaustive] so new block kinds can be added without a breaking change; only Text, ToolUse, and ToolResult are exercised in v0.

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

Text

Plain text.

Fields

text: String

The text content.

Image

An image (multimodal).

Fields

source: ImageSource

Where the image bytes come from.

Reasoning

Assistant reasoning, unified across wires (ADR-0013 amendment 2026-07-19; replaces the earlier Thinking/RedactedThinking pair).

ReasoningFormat selects the replay contract; each wire’s build replays only its own format(s) and drops foreign formats (a session never crosses wires, so nothing is lost).

Fields

format: ReasoningFormat

Which encoding/replay contract this data follows.

text: String

Human-readable reasoning: the full text (anthropic), empty (anthropic_redacted), the summary (openai_responses), or the captured text (text_only).

signature: Option<String>

Anthropic’s validator over text (anthropic format only).

payload: Option<Value>

The wire’s opaque replay payload, replayed verbatim and never interpreted: the whole Responses reasoning item (openai_responses) or Anthropic’s redacted-thinking data (anthropic_redacted).

ToolUse

A tool call emitted by the assistant.

Fields

id: String

Provider-assigned id, paired with a later ContentBlock::ToolResult.

name: String

The client-facing tool name (the harness pack’s name).

input: Value

The tool arguments as a JSON value.

ToolResult

The result of a tool call, carried in a Role::User message.

Fields

tool_use_id: String

The id of the ContentBlock::ToolUse this answers.

content: Vec<ResultChunk>

The result content (text and/or images).

is_error: bool

Whether the tool call failed (a soft error the model can recover from).

Trait Implementations

impl Clone for ContentBlock

fn clone(&self) -> ContentBlock

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ContentBlock

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ContentBlock

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for ContentBlock

fn eq(&self, other: &ContentBlock) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for ContentBlock

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for ContentBlock