Struct MetadataContextAttributionResultContextAttribution 

pub struct MetadataContextAttributionResultContextAttribution {
    pub buffer_tokens: i64,
    pub categories: MetadataContextAttributionResultContextAttributionCategories,
    pub compactions: MetadataContextAttributionResultContextAttributionCompactions,
    pub compaction_threshold: i64,
    pub entries: Vec<MetadataContextAttributionResultContextAttributionEntriesItem>,
    pub limit: i64,
    pub model_id: String,
    pub model_source: String,
    pub prompt_token_limit: i64,
    pub total_tokens: i64,
}

Per-source token attribution snapshot for the current context window. The heaviest individual messages are available separately via metadata.getContextHeaviestMessages.

Fields

buffer_tokens: i64

Output reserve plus the tokens past the buffer-exhaustion blocking threshold. Mirrors SessionContextInfo.bufferTokens.

categories: MetadataContextAttributionResultContextAttributionCategories

The six normalized /context header buckets, computed from the same tokenization as entries so the two never disagree. Convenience rollups: freeSpace and buffer describe window capacity rather than occupied context, so the values do not sum to totalTokens.

compactions: MetadataContextAttributionResultContextAttributionCompactions

Successful compaction history for the session.

compaction_threshold: i64

Token count at which background compaction starts. Mirrors SessionContextInfo.compactionThreshold.

entries: Vec<MetadataContextAttributionResultContextAttributionEntriesItem>

Flat list of per-source attribution entries. Group by kind and render unrecognized kinds generically. Nesting and rollups are expressed via parentId.

limit: i64

Prompt limit plus the model’s output reserve: the full context window categories.freeSpace and categories.buffer are measured against. Mirrors SessionContextInfo.limit.

model_id: String

The concrete model id the entire breakdown was tokenized against (feeds the per-model token multiplier). Under Auto (Free/Student) this is the resolved model, not the literal auto sentinel, so totals are not undercounted. A single-model approximation of a potentially multi-model Auto session.

model_source: String

How modelId was chosen. Not a closed set — tolerate unknown values. Known values today: autoResolved (the model Auto resolved to), selected (the user’s explicitly selected model), default (a fallback before any model is known).

prompt_token_limit: i64

Maximum prompt tokens the resolved model accepts — the denominator for a ##k/###k context-usage display. Mirrors SessionContextInfo.promptTokenLimit.

total_tokens: i64

Total token count of the current context window the entries are measured against (system message + conversation messages + tool definitions — the same total reported by /context). Divide an entry’s tokens by this to derive its share.

Trait Implementations

impl Clone for MetadataContextAttributionResultContextAttribution

fn clone(&self) -> MetadataContextAttributionResultContextAttribution

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MetadataContextAttributionResultContextAttribution

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

Formats the value using the given formatter.

impl Default for MetadataContextAttributionResultContextAttribution

fn default() -> MetadataContextAttributionResultContextAttribution

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for MetadataContextAttributionResultContextAttribution

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for MetadataContextAttributionResultContextAttribution

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

where __S: Serializer,

Serialize this value into the given Serde serializer.