Enum ModelEntry 

pub enum ModelEntry {
    Id(String),
    Detailed {
        id: String,
        name: Option<String>,
        context_window: Option<u64>,
        max_output_tokens: Option<u64>,
        thinking_format: Option<ThinkingFormat>,
    },
}

A model candidate configuration declaration.

TOML supports two forms (accepted via Deserialize with untagged):

• Plain string "gpt-5.5": only the id is given; the UI display name falls back to the id.
• Table { id = "...", name = "Opus 4.8" }: pairs a long id with a short display name.

name is mapped to defect_agent::llm::ModelInfo::display_name; the ACP uses it as the label for model selector options. When None, the wire layer falls back to the id.

context_window / max_output_tokens let the user declare model metadata that the provider cannot discover at runtime — most importantly for Bedrock, whose SDK does not return model limits, so without this the compaction watermarks have no window to key off and the context can grow unbounded. Both are optional; when absent the provider’s own value (if any) or the compaction fallback applies.

Variants

Id(String)

A plain ID (legacy format).

Detailed

A detailed entry with a display name and optional limits.

Fields

id: String

name: Option<String>

context_window: Option<u64>

max_output_tokens: Option<u64>

thinking_format: Option<ThinkingFormat>

How extended thinking is serialized on the Anthropic Messages wire for this model. Defaults to ThinkingFormat::Adaptive — the format newer models (Opus 4.6+, Sonnet 4.6, Fable) require. Set to ThinkingFormat::Legacy for older models (Sonnet 4.5, Haiku 4.5, …) that only accept the thinking.type=enabled + budget_tokens shape and reject adaptive / output_config.effort. Ignored by the OpenAI-compatible protocol.

Implementations

impl ModelEntry

pub fn id(&self) -> &str

The model ID (present in both variants).

pub fn name(&self) -> Option<&str>

Optional display name (only available in table form).

pub fn context_window(&self) -> Option<u64>

Optional context-window size (tokens), only from the table form.

pub fn max_output_tokens(&self) -> Option<u64>

Optional max output tokens, only from the table form.

pub fn thinking_format(&self) -> Option<ThinkingFormat>

Optional thinking wire format override, only from the table form.

Trait Implementations

impl Clone for ModelEntry

fn clone(&self) -> ModelEntry

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ModelEntry

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ModelEntry

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for ModelEntry

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for ModelEntry