@kreuzberg/liter-llm-node 1.6.1 → 1.6.3

package/index.d.ts

@@ -1,10 +1,16 @@
 // This file is auto-generated by alef — DO NOT EDIT.
-// alef:hash:9ebda2bedb27bc07d8a4cfcbfadd55fd159de7f875fcb63308eaee6e1c3a895d
+// alef:hash:0c694843abf5c6a9fc8b410dfa73aea7c94ee538a40dc746dd72c50e7c986f97
 // To regenerate: alef generate
 // To verify freshness: alef verify --exit-code
 /* eslint-disable */
 
-export type JsonValue = string | number | boolean | null | JsonValue[] | { [key: string]: JsonValue };
+export type JsonValue =
+  | string
+  | number
+  | boolean
+  | null
+  | JsonValue[]
+  | { [key: string]: JsonValue };
 
 /**
  * Return all provider configs from the registry.
@@ -19,9 +25,8 @@ export declare function allProviders(): Array<ProviderConfig>;
  * Return the capability flags for a named provider.
  *
  * Performs an O(n) linear scan over the embedded registry (143 entries).
- * Returns an owned value so that bindings can box/copy it across the FFI
- * boundary without dealing with lifetimes. `ProviderCapabilities` is `Copy`,
- * so this is a cheap memcpy of seven `bool` fields.
+ * Returns an owned value so bindings can pass capability data without
+ * borrowing registry internals.
  *
  * For unknown `provider_name` values the function returns an all-`false`
  * sentinel so callers never need to handle `Option`.
@@ -35,7 +40,12 @@ export declare function capabilities(providerName: string): ProviderCapabilities
  * stay below `limit`.  Returns `Err(LiterLlmError::Streaming)` on overflow
  * and emits a `tracing::warn!` with context.
  */
-export declare function checkBound(context: string, currentLen: number, incoming: number, limit: number): void;
+export declare function checkBound(
+  context: string,
+  currentLen: number,
+  incoming: number,
+  limit: number,
+): void;
 
 /**
  * Remove all guardrails from the global registry.
@@ -55,7 +65,11 @@ export declare function clear(): void;
  * are tried by stripping from the last `-` or `.` separator.  For example,
  * `gpt-4-0613` will match `gpt-4` if no `gpt-4-0613` entry exists.
  */
-export declare function completionCost(model: string, promptTokens: number, completionTokens: number): number | null;
+export declare function completionCost(
+  model: string,
+  promptTokens: number,
+  completionTokens: number,
+): number | null;
 
 /**
  * Calculate the estimated cost of a completion, accounting for cached
@@ -71,7 +85,12 @@ export declare function completionCost(model: string, promptTokens: number, comp
  * Returns `None` if the model is not present in the embedded pricing
  * registry, mirroring [`completion_cost`].
  */
-export declare function completionCostWithCache(model: string, promptTokens: number, cachedTokens: number, completionTokens: number): number | null;
+export declare function completionCostWithCache(
+  model: string,
+  promptTokens: number,
+  cachedTokens: number,
+  completionTokens: number,
+): number | null;
 
 /**
  * Return the set of complex provider names.
@@ -93,7 +112,10 @@ export declare function complexProviderNames(): Array<string>;
  * @throws Returns [`LiterLlmError::BadRequest`] if the tokenizer cannot be loaded or
  * if tokenization fails for any message.
  */
-export declare function countRequestTokens(model: string, req?: ChatCompletionRequest | undefined | null): number;
+export declare function countRequestTokens(
+  model: string,
+  req?: ChatCompletionRequest | undefined | null,
+): number;
 
 /**
  * Count tokens in a text string using the tokenizer for the given model.
@@ -115,7 +137,13 @@ export declare function countTokens(model: string, text: string): number;
  * @throws Returns [`LiterLlmError`] if the underlying HTTP client cannot be
  * constructed, or if the resolved provider configuration is invalid.
  */
-export declare function createClient(apiKey: string, baseUrl?: string | undefined | null, timeoutSecs?: number | undefined | null, maxRetries?: number | undefined | null, modelHint?: string | undefined | null): DefaultClient;
+export declare function createClient(
+  apiKey: string,
+  baseUrl?: string | undefined | null,
+  timeoutSecs?: number | undefined | null,
+  maxRetries?: number | undefined | null,
+  modelHint?: string | undefined | null,
+): DefaultClient;
 
 /**
  * Create a new LLM client from a JSON string.
@@ -130,8 +158,8 @@ export declare function createClientFromJson(json: string): DefaultClient;
  * Install the `ring` crypto provider as the rustls process default, idempotently.
  *
  * rustls 0.23+ removed the implicit default provider. This function installs
- * `ring` once per process. Subsequent calls are no-ops. Calling it from a
- * downstream Rust app that has already installed `aws-lc-rs` is safe — the
+ * `ring` once per process. Subsequent calls are no-ops. Calling it after
+ * another rustls crypto provider has already been installed is safe: the
  * `Err` from `install_default()` is silently ignored.
  *
  * Called automatically by every internal `reqwest::Client` constructor
@@ -149,34 +177,34 @@ export declare function ensureCryptoProvider(): void;
 /** Assistant's response to a user message. */
 export interface AssistantMessage {
   /** The assistant's text response. Absent if tool calls are returned instead. */
-  readonly content?: string
+  readonly content?: string;
   /** Optional name for the assistant. */
-  readonly name?: string
+  readonly name?: string;
   /** Tool calls the model wants to execute, if any. */
-  readonly toolCalls?: Array<ToolCall>
+  readonly toolCalls?: Array<ToolCall>;
   /** Refusal reason, if the model declined to respond per safety policies. */
-  readonly refusal?: string
+  readonly refusal?: string;
   /** Deprecated legacy function_call field; retained for API compatibility. */
-  readonly functionCall?: FunctionCall
+  readonly functionCall?: FunctionCall;
 }
 
 /** Audio content part for speech-capable models. */
 export interface AudioContent {
   /** Base64-encoded audio data. */
-  readonly data?: string
+  readonly data?: string;
   /** Audio format (e.g., "wav", "mp3", "ogg"). */
-  readonly format?: string
+  readonly format?: string;
 }
 
 /** Auth configuration block. */
 export interface AuthConfig {
   /** Auth scheme classification. */
-  readonly authType: AuthType
+  readonly authType: AuthType;
   /**
    * Name of the environment variable that holds the API key (e.g. `"OPENAI_API_KEY"`).
    * Holds the variable name, never the secret value.
    */
-  readonly envVar?: string
+  readonly envVar?: string;
 }
 
 /** How the API key is sent in the HTTP request. */
@@ -204,65 +232,65 @@ export declare enum AuthType {
 /** Query parameters for listing batches. */
 export interface BatchListQuery {
   /** Maximum number of results to return. Defaults to 20. */
-  readonly limit?: number
+  readonly limit?: number;
   /** Pagination cursor: return results after this batch ID. */
-  readonly after?: string
+  readonly after?: string;
 }
 
 /** Response from listing batches. */
 export interface BatchListResponse {
   /** Object type (always `"list"`). */
-  readonly object?: string
+  readonly object?: string;
   /** List of batch objects. */
-  readonly data?: Array<BatchObject>
+  readonly data?: Array<BatchObject>;
   /** Whether more results are available. */
-  readonly hasMore?: boolean
+  readonly hasMore?: boolean;
   /** First batch ID in the result set (for pagination). */
-  readonly firstId?: string
+  readonly firstId?: string;
   /** Last batch ID in the result set (for pagination). */
-  readonly lastId?: string
+  readonly lastId?: string;
 }
 
 /** A batch job object. */
 export interface BatchObject {
   /** Unique batch ID. */
-  readonly id?: string
+  readonly id?: string;
   /** Object type (always `"batch"`). */
-  readonly object?: string
+  readonly object?: string;
   /** API endpoint (e.g., `"/v1/chat/completions"`). */
-  readonly endpoint?: string
+  readonly endpoint?: string;
   /** ID of the input file. */
-  readonly inputFileId?: string
+  readonly inputFileId?: string;
   /** Completion window (e.g., `"24h"`). */
-  readonly completionWindow?: string
+  readonly completionWindow?: string;
   /** Current job status. */
-  readonly status?: BatchStatus
+  readonly status?: BatchStatus;
   /** ID of the output file (present when completed). */
-  readonly outputFileId?: string
+  readonly outputFileId?: string;
   /** ID of the error file (present if some requests failed). */
-  readonly errorFileId?: string
+  readonly errorFileId?: string;
   /** Unix timestamp of batch creation. */
-  readonly createdAt?: number
+  readonly createdAt?: number;
   /** Unix timestamp of completion (if completed). */
-  readonly completedAt?: number
+  readonly completedAt?: number;
   /** Unix timestamp of failure (if failed). */
-  readonly failedAt?: number
+  readonly failedAt?: number;
   /** Unix timestamp of expiration (if expired). */
-  readonly expiredAt?: number
+  readonly expiredAt?: number;
   /** Request processing counts. */
-  readonly requestCounts?: BatchRequestCounts
+  readonly requestCounts?: BatchRequestCounts;
   /** Metadata attached to the batch. */
-  readonly metadata?: JsonValue
+  readonly metadata?: JsonValue;
 }
 
 /** Request processing counts for a batch. */
 export interface BatchRequestCounts {
   /** Total requests in the batch. */
-  readonly total?: number
+  readonly total?: number;
   /** Completed requests. */
-  readonly completed?: number
+  readonly completed?: number;
   /** Failed requests. */
-  readonly failed?: number
+  readonly failed?: number;
 }
 
 /** Status of a batch job. */
@@ -288,147 +316,147 @@ export declare enum BatchStatus {
 /** Configuration for budget enforcement. */
 export interface BudgetConfig {
   /** Maximum total spend across all models, in USD.  `None` means unlimited. */
-  readonly globalLimit?: number
+  readonly globalLimit?: number;
   /**
    * Per-model spending limits in USD.  Models not listed here are only
    * constrained by `global_limit`.
    */
-  readonly modelLimits?: Record<string, number>
+  readonly modelLimits?: Record<string, number>;
   /** Whether to reject requests or merely warn when a limit is exceeded. */
-  readonly enforcement?: Enforcement
+  readonly enforcement?: Enforcement;
 }
 
 /** Storage backend for the response cache. */
 export type CacheBackend =
-  | { type: 'memory' }
-  | { type: 'open_dal'; scheme: string; config: Record<string, string> }
+  | { type: "memory" }
+  | { type: "open_dal"; scheme: string; config: Record<string, string> };
 
 /** Configuration for the response cache. */
 export interface CacheConfig {
   /** Maximum number of cached entries. */
-  readonly maxEntries?: number
+  readonly maxEntries?: number;
   /** Time-to-live for each cached entry. */
-  readonly ttl?: number
+  readonly ttl?: number;
   /** Storage backend to use. */
-  readonly backend?: CacheBackend
+  readonly backend?: CacheBackend;
 }
 
 /** A streamed chunk of a chat completion response. */
 export interface ChatCompletionChunk {
   /** Unique identifier for this stream. */
-  readonly id?: string
+  readonly id?: string;
   /**
    * Always `"chat.completion.chunk"` from OpenAI-compatible APIs.  Stored
    * as a plain `String` so non-standard provider values do not fail parsing.
    */
-  readonly object?: string
+  readonly object?: string;
   /** Unix timestamp of chunk creation. */
-  readonly created?: number
+  readonly created?: number;
   /** Model used to generate the chunk. */
-  readonly model?: string
+  readonly model?: string;
   /** Streaming choices (delta updates). */
-  readonly choices?: Array<StreamChoice>
+  readonly choices?: Array<StreamChoice>;
   /** Token usage (typically only in the final chunk). */
-  readonly usage?: Usage
+  readonly usage?: Usage;
   /** Fingerprint of the system configuration (OpenAI-specific). */
-  readonly systemFingerprint?: string
+  readonly systemFingerprint?: string;
   /** Service tier used (OpenAI-specific). */
-  readonly serviceTier?: string
+  readonly serviceTier?: string;
 }
 
 /** Chat completion request (compatible with OpenAI and similar APIs). */
 export interface ChatCompletionRequest {
   /** Model ID (e.g., `"gpt-4o-mini"`, `"claude-3-5-sonnet"`). */
-  readonly model?: string
+  readonly model?: string;
   /** Conversation history from oldest to newest. */
-  readonly messages?: Array<Message>
+  readonly messages?: Array<Message>;
   /** Sampling temperature in `[0.0, 2.0]`. Higher increases randomness. Defaults to 1.0. */
-  readonly temperature?: number
+  readonly temperature?: number;
   /** Nucleus sampling parameter in `[0.0, 1.0]`. Lower is more focused. */
-  readonly topP?: number
+  readonly topP?: number;
   /** Number of chat completions to generate. Defaults to 1. */
-  readonly n?: number
+  readonly n?: number;
   /**
    * Whether to stream the response.
    *
    * Managed by the client layer — do not set directly.
    */
-  readonly stream?: boolean
+  readonly stream?: boolean;
   /** Stop sequence(s) that halt token generation. */
-  readonly stop?: StopSequence
+  readonly stop?: StopSequence;
   /** Max output tokens. Different from max_completion_tokens in some providers. */
-  readonly maxTokens?: number
+  readonly maxTokens?: number;
   /** Presence penalty in `[-2.0, 2.0]`. Positive discourages repeated topics. */
-  readonly presencePenalty?: number
+  readonly presencePenalty?: number;
   /** Frequency penalty in `[-2.0, 2.0]`. Positive discourages repeated tokens. */
-  readonly frequencyPenalty?: number
+  readonly frequencyPenalty?: number;
   /**
    * Token bias map.  Uses `BTreeMap` (sorted keys) for deterministic
    * serialization order — important when hashing or signing requests.
    */
-  readonly logitBias?: Record<string, number>
+  readonly logitBias?: Record<string, number>;
   /** User identifier for request tracking and abuse detection. */
-  readonly user?: string
+  readonly user?: string;
   /** Tools the model can invoke. */
-  readonly tools?: Array<ChatCompletionTool>
+  readonly tools?: Array<ChatCompletionTool>;
   /** Tool usage mode (auto, required, none, or specific tool). */
-  readonly toolChoice?: ToolChoice
+  readonly toolChoice?: ToolChoice;
   /** Whether the model can call multiple tools in parallel. Defaults to true. */
-  readonly parallelToolCalls?: boolean
+  readonly parallelToolCalls?: boolean;
   /** Output format constraint (text, JSON, JSON schema). */
-  readonly responseFormat?: ResponseFormat
+  readonly responseFormat?: ResponseFormat;
   /** Streaming options (e.g., include_usage). */
-  readonly streamOptions?: StreamOptions
+  readonly streamOptions?: StreamOptions;
   /** Random seed for reproducible outputs. Provider support varies. */
-  readonly seed?: number
+  readonly seed?: number;
   /** Reasoning effort level (low, medium, high) for extended-thinking models. */
-  readonly reasoningEffort?: ReasoningEffort
+  readonly reasoningEffort?: ReasoningEffort;
   /**
    * Provider-specific extra parameters merged into the request body.
    * Use for guardrails, safety settings, grounding config, etc.
    */
-  readonly extraBody?: JsonValue
+  readonly extraBody?: JsonValue;
 }
 
 /** Chat completion response from the API. */
 export interface ChatCompletionResponse {
   /** Unique identifier for this response. */
-  readonly id?: string
+  readonly id?: string;
   /**
    * Always `"chat.completion"` from OpenAI-compatible APIs.  Stored as a
    * plain `String` so non-standard provider values do not break deserialization.
    */
-  readonly object?: string
+  readonly object?: string;
   /** Unix timestamp of response creation. */
-  readonly created?: number
+  readonly created?: number;
   /** Model used to generate the response. */
-  readonly model?: string
+  readonly model?: string;
   /** List of completion choices. */
-  readonly choices?: Array<Choice>
+  readonly choices?: Array<Choice>;
   /** Token usage statistics. */
-  readonly usage?: Usage
+  readonly usage?: Usage;
   /** Fingerprint of the system configuration (OpenAI-specific). */
-  readonly systemFingerprint?: string
+  readonly systemFingerprint?: string;
   /** Service tier used (OpenAI-specific). */
-  readonly serviceTier?: string
+  readonly serviceTier?: string;
 }
 
 /** A tool the model can invoke (currently, all tools are functions). */
 export interface ChatCompletionTool {
   /** Tool type (always "function" in OpenAI spec). */
-  readonly toolType: ToolType
+  readonly toolType: ToolType;
   /** Function definition with name, description, and JSON schema parameters. */
-  readonly function: FunctionDefinition
+  readonly function: FunctionDefinition;
 }
 
 /** A single completion choice. */
 export interface Choice {
   /** Index of this choice in the choices array. */
-  readonly index?: number
+  readonly index?: number;
   /** The assistant's message response. */
-  readonly message?: AssistantMessage
+  readonly message?: AssistantMessage;
   /** Why the model stopped generating (stop, length, tool_calls, content_filter, etc.). */
-  readonly finishReason?: FinishReason
+  readonly finishReason?: FinishReason;
 }
 
 /**
@@ -438,8 +466,8 @@ export interface Choice {
  * to pass it through (optionally modified), `Ok(None)` to drop the chunk,
  * or `Err(e)` to propagate a stream error.
  *
- * The trait is object-safe so implementations can be stored in a
- * `Vec<Box<dyn ChunkMiddleware>>` inside [`StreamPipeline`].
+ * The trait is object-safe so multiple middleware implementations can be
+ * chained inside [`StreamPipeline`].
  */
 export interface ChunkMiddleware {
   /**
@@ -449,7 +477,7 @@ export interface ChunkMiddleware {
    * - `Ok(None)` — drop this chunk silently.
    * - `Err(e)` — propagate as a stream error.
    */
-  process(chunk?: ChatCompletionChunk | undefined | null): string
+  process(chunk?: ChatCompletionChunk | undefined | null): string;
 }
 
 /** Observable state of a circuit breaker. */
@@ -464,111 +492,111 @@ export declare enum CircuitState {
 
 /** A single content part in a user message — text, image, document, or audio. */
 export type ContentPart =
-  | { type: 'text'; text: string }
-  | { type: 'image_url'; imageUrl: ImageUrl }
-  | { type: 'document'; document: DocumentContent }
-  | { type: 'input_audio'; inputAudio: AudioContent }
+  | { type: "text"; text: string }
+  | { type: "image_url"; imageUrl: ImageUrl }
+  | { type: "document"; document: DocumentContent }
+  | { type: "input_audio"; inputAudio: AudioContent };
 
 /** Request to create a batch job. */
 export interface CreateBatchRequest {
   /** ID of the uploaded input file (JSONL format). */
-  readonly inputFileId?: string
+  readonly inputFileId?: string;
   /** API endpoint (e.g., `"/v1/chat/completions"`). */
-  readonly endpoint?: string
+  readonly endpoint?: string;
   /** Completion window (e.g., `"24h"`). */
-  readonly completionWindow?: string
+  readonly completionWindow?: string;
   /** Optional metadata to attach to the batch. */
-  readonly metadata?: JsonValue
+  readonly metadata?: JsonValue;
 }
 
 /** Request to upload a file. */
 export interface CreateFileRequest {
   /** Base64-encoded file data. */
-  readonly file?: string
+  readonly file?: string;
   /** Purpose for the file. */
-  readonly purpose?: FilePurpose
+  readonly purpose?: FilePurpose;
   /** Optional filename to associate with the upload. */
-  readonly filename?: string
+  readonly filename?: string;
 }
 
 /** Request to create images from a text prompt. */
 export interface CreateImageRequest {
   /** Text description of the image to generate. */
-  readonly prompt?: string
+  readonly prompt?: string;
   /** Model ID (e.g., `"dall-e-3"`). Optional; API may use default if unset. */
-  readonly model?: string
+  readonly model?: string;
   /** Number of images to generate. Defaults to 1. */
-  readonly n?: number
+  readonly n?: number;
   /** Image size (e.g., `"1024x1024"`, `"1792x1024"`). */
-  readonly size?: string
+  readonly size?: string;
   /** Image quality: `"standard"` or `"hd"`. */
-  readonly quality?: string
+  readonly quality?: string;
   /** Style: `"natural"` or `"vivid"` (DALL-E 3 only). */
-  readonly style?: string
+  readonly style?: string;
   /** Response format: `"url"` or `"b64_json"`. */
-  readonly responseFormat?: string
+  readonly responseFormat?: string;
   /** User identifier for request tracking. */
-  readonly user?: string
+  readonly user?: string;
 }
 
 /** Request to create a structured response. */
 export interface CreateResponseRequest {
   /** Model ID. */
-  readonly model?: string
+  readonly model?: string;
   /** Input data to process (e.g., a document to extract from). */
-  readonly input?: JsonValue
+  readonly input?: JsonValue;
   /** Instructions for processing the input. */
-  readonly instructions?: string
+  readonly instructions?: string;
   /** Available tools the model can use. */
-  readonly tools?: Array<ResponseTool>
+  readonly tools?: Array<ResponseTool>;
   /** Sampling temperature in `[0.0, 2.0]`. Defaults to 1.0. */
-  readonly temperature?: number
+  readonly temperature?: number;
   /** Maximum output tokens. */
-  readonly maxOutputTokens?: number
+  readonly maxOutputTokens?: number;
   /** Optional metadata. */
-  readonly metadata?: JsonValue
+  readonly metadata?: JsonValue;
 }
 
 /** Request to generate speech audio from text. */
 export interface CreateSpeechRequest {
   /** Model ID (e.g., `"tts-1"`, `"tts-1-hd"`). */
-  readonly model?: string
+  readonly model?: string;
   /** Text to synthesize into speech. */
-  readonly input?: string
+  readonly input?: string;
   /** Voice name (e.g., `"alloy"`, `"echo"`, `"fable"`, `"onyx"`, `"nova"`, `"shimmer"`). */
-  readonly voice?: string
+  readonly voice?: string;
   /** Audio format (e.g., `"mp3"`, `"opus"`, `"aac"`, `"flac"`, `"wav"`, `"pcm"`). */
-  readonly responseFormat?: string
+  readonly responseFormat?: string;
   /** Playback speed in `[0.25, 4.0]`. Defaults to 1.0. */
-  readonly speed?: number
+  readonly speed?: number;
 }
 
 /** Request to transcribe audio into text. */
 export interface CreateTranscriptionRequest {
   /** Model ID (e.g., `"whisper-1"`). */
-  readonly model?: string
+  readonly model?: string;
   /** Base64-encoded audio file data. */
-  readonly file?: string
+  readonly file?: string;
   /** Language ISO-639-1 code (e.g., `"en"`, `"fr"`, `"de"`). Optional; model auto-detects. */
-  readonly language?: string
+  readonly language?: string;
   /** Optional text to guide the model (improves accuracy for domain-specific terms). */
-  readonly prompt?: string
+  readonly prompt?: string;
   /** Output format (e.g., `"json"`, `"text"`, `"vtt"`, `"srt"`, `"verbose_json"`). */
-  readonly responseFormat?: string
+  readonly responseFormat?: string;
   /** Sampling temperature in `[0.0, 1.0]`. Higher increases variability. Defaults to 0. */
-  readonly temperature?: number
+  readonly temperature?: number;
 }
 
 /** Configuration for registering a custom LLM provider at runtime. */
 export interface CustomProviderConfig {
   /** Unique name for this provider (e.g., "my-provider"). */
-  readonly name: string
+  readonly name: string;
   /** Base URL for the provider's API (e.g., "https://api.my-provider.com/v1"). */
-  readonly baseUrl: string
+  readonly baseUrl: string;
   /** Authentication header format. */
-  readonly authHeader: AuthHeaderFormat
+  readonly authHeader: AuthHeaderFormat;
   /** Model name prefixes that route to this provider (e.g., `["my-"]`). */
-  readonly modelPrefixes: Array<string>
+  readonly modelPrefixes: Array<string>;
 }
 
 /**
@@ -589,27 +617,29 @@ export interface CustomProviderConfig {
  * headers are cached at construction to avoid redundant encoding on every request.
  */
 export declare class DefaultClient {
-  chat(req?: ChatCompletionRequest | undefined | null): Promise<ChatCompletionResponse>
-  chatStream(req?: ChatCompletionRequest | undefined | null): Promise<AsyncGenerator<ChatCompletionChunk, void, undefined>>
-  embed(req?: EmbeddingRequest | undefined | null): Promise<EmbeddingResponse>
-  listModels(): Promise<ModelsListResponse>
-  imageGenerate(req?: CreateImageRequest | undefined | null): Promise<ImagesResponse>
-  speech(req?: CreateSpeechRequest | undefined | null): Promise<Uint8Array>
-  transcribe(req?: CreateTranscriptionRequest | undefined | null): Promise<TranscriptionResponse>
-  moderate(req?: ModerationRequest | undefined | null): Promise<ModerationResponse>
-  rerank(req?: RerankRequest | undefined | null): Promise<RerankResponse>
-  search(req?: SearchRequest | undefined | null): Promise<SearchResponse>
-  ocr(req?: OcrRequest | undefined | null): Promise<OcrResponse>
-  createFile(req?: CreateFileRequest | undefined | null): Promise<FileObject>
-  retrieveFile(fileId: string): Promise<FileObject>
-  deleteFile(fileId: string): Promise<DeleteResponse>
-  listFiles(query?: FileListQuery | undefined | null): Promise<FileListResponse>
-  fileContent(fileId: string): Promise<Uint8Array>
-  createBatch(req?: CreateBatchRequest | undefined | null): Promise<BatchObject>
-  retrieveBatch(batchId: string): Promise<BatchObject>
-  listBatches(query?: BatchListQuery | undefined | null): Promise<BatchListResponse>
-  cancelBatch(batchId: string): Promise<BatchObject>
-  fetchBatchForPolling(batchId: string): Promise<BatchObject>
+  chat(req?: ChatCompletionRequest | undefined | null): Promise<ChatCompletionResponse>;
+  chatStream(
+    req?: ChatCompletionRequest | undefined | null,
+  ): Promise<AsyncGenerator<ChatCompletionChunk, void, undefined>>;
+  embed(req?: EmbeddingRequest | undefined | null): Promise<EmbeddingResponse>;
+  listModels(): Promise<ModelsListResponse>;
+  imageGenerate(req?: CreateImageRequest | undefined | null): Promise<ImagesResponse>;
+  speech(req?: CreateSpeechRequest | undefined | null): Promise<Uint8Array>;
+  transcribe(req?: CreateTranscriptionRequest | undefined | null): Promise<TranscriptionResponse>;
+  moderate(req?: ModerationRequest | undefined | null): Promise<ModerationResponse>;
+  rerank(req?: RerankRequest | undefined | null): Promise<RerankResponse>;
+  search(req?: SearchRequest | undefined | null): Promise<SearchResponse>;
+  ocr(req?: OcrRequest | undefined | null): Promise<OcrResponse>;
+  createFile(req?: CreateFileRequest | undefined | null): Promise<FileObject>;
+  retrieveFile(fileId: string): Promise<FileObject>;
+  deleteFile(fileId: string): Promise<DeleteResponse>;
+  listFiles(query?: FileListQuery | undefined | null): Promise<FileListResponse>;
+  fileContent(fileId: string): Promise<Uint8Array>;
+  createBatch(req?: CreateBatchRequest | undefined | null): Promise<BatchObject>;
+  retrieveBatch(batchId: string): Promise<BatchObject>;
+  listBatches(query?: BatchListQuery | undefined | null): Promise<BatchListResponse>;
+  cancelBatch(batchId: string): Promise<BatchObject>;
+  fetchBatchForPolling(batchId: string): Promise<BatchObject>;
   /**
    * Poll a batch until it reaches a terminal status (Completed, Failed, Expired, Cancelled).
    *
@@ -619,36 +649,39 @@ export declare class DefaultClient {
    * Returns `BatchWaitError::Timeout` if the configured timeout is exceeded.
    * Returns `BatchWaitError::Client` for underlying client errors.
    */
-  waitForBatch(batchId: string, config?: WaitForBatchConfig | undefined | null): Promise<BatchObject>
-  createResponse(req?: CreateResponseRequest | undefined | null): Promise<ResponseObject>
-  retrieveResponse(responseId: string): Promise<ResponseObject>
-  cancelResponse(responseId: string): Promise<ResponseObject>
+  waitForBatch(
+    batchId: string,
+    config?: WaitForBatchConfig | undefined | null,
+  ): Promise<BatchObject>;
+  createResponse(req?: CreateResponseRequest | undefined | null): Promise<ResponseObject>;
+  retrieveResponse(responseId: string): Promise<ResponseObject>;
+  cancelResponse(responseId: string): Promise<ResponseObject>;
 }
 
 /** Response from a delete operation. */
 export interface DeleteResponse {
   /** ID of the deleted resource. */
-  readonly id?: string
+  readonly id?: string;
   /** Object type. */
-  readonly object?: string
+  readonly object?: string;
   /** Confirmation that the resource was deleted. */
-  readonly deleted?: boolean
+  readonly deleted?: boolean;
 }
 
 /** Developer message (system-like message for Claude models). */
 export interface DeveloperMessage {
   /** Developer-specific instructions or context. */
-  readonly content?: string
+  readonly content?: string;
   /** Optional name for the developer message source. */
-  readonly name?: string
+  readonly name?: string;
 }
 
 /** PDF/document content part for vision-capable models. */
 export interface DocumentContent {
   /** Base64-encoded document data or URL. */
-  readonly data?: string
+  readonly data?: string;
   /** MIME type (e.g., "application/pdf", "text/csv"). */
-  readonly mediaType?: string
+  readonly mediaType?: string;
 }
 
 /** The format in which the embedding vectors are returned. */
@@ -673,25 +706,25 @@ export interface EmbeddingObject {
    * Always `"embedding"` from OpenAI-compatible APIs.  Stored as a plain
    * `String` so non-standard provider values do not break deserialization.
    */
-  readonly object: string
+  readonly object: string;
   /** The embedding vector. */
-  readonly embedding: Array<number>
+  readonly embedding: Array<number>;
   /** Index in the batch (corresponds to input order). */
-  readonly index: number
+  readonly index: number;
 }
 
 /** Embedding request. */
 export interface EmbeddingRequest {
   /** Model ID (e.g., `"text-embedding-3-small"`). */
-  readonly model?: string
+  readonly model?: string;
   /** Text or texts to embed. */
-  readonly input?: EmbeddingInput
+  readonly input?: EmbeddingInput;
   /** Output format: float (native) or base64. */
-  readonly encodingFormat?: EmbeddingFormat
+  readonly encodingFormat?: EmbeddingFormat;
   /** Requested embedding dimensions (if supported by the model). */
-  readonly dimensions?: number
+  readonly dimensions?: number;
   /** User identifier for request tracking. */
-  readonly user?: string
+  readonly user?: string;
 }
 
 /** Embedding response. */
@@ -700,13 +733,13 @@ export interface EmbeddingResponse {
    * Always `"list"` from OpenAI-compatible APIs.  Stored as a plain
    * `String` so non-standard provider values do not break deserialization.
    */
-  readonly object: string
+  readonly object: string;
   /** List of embeddings. */
-  readonly data: Array<EmbeddingObject>
+  readonly data: Array<EmbeddingObject>;
   /** Model used to generate embeddings. */
-  readonly model: string
+  readonly model: string;
   /** Token usage (input tokens only; embeddings have zero output tokens). */
-  readonly usage?: Usage
+  readonly usage?: Usage;
 }
 
 /** How budget limits are enforced. */
@@ -726,39 +759,39 @@ export declare enum Enforcement {
 /** Query parameters for listing files. */
 export interface FileListQuery {
   /** Filter by file purpose (e.g., `"batch"`, `"fine-tune"`). */
-  readonly purpose?: string
+  readonly purpose?: string;
   /** Maximum number of results to return. Defaults to 20. */
-  readonly limit?: number
+  readonly limit?: number;
   /** Pagination cursor: return results after this file ID. */
-  readonly after?: string
+  readonly after?: string;
 }
 
 /** Response from listing files. */
 export interface FileListResponse {
   /** Object type (always `"list"`). */
-  readonly object?: string
+  readonly object?: string;
   /** List of file objects. */
-  readonly data?: Array<FileObject>
+  readonly data?: Array<FileObject>;
   /** Whether more results are available. */
-  readonly hasMore?: boolean
+  readonly hasMore?: boolean;
 }
 
 /** An uploaded file object. */
 export interface FileObject {
   /** Unique file ID. */
-  readonly id?: string
+  readonly id?: string;
   /** Object type (always `"file"`). */
-  readonly object?: string
+  readonly object?: string;
   /** File size in bytes. */
-  readonly bytes?: number
+  readonly bytes?: number;
   /** Unix timestamp of file creation. */
-  readonly createdAt?: number
+  readonly createdAt?: number;
   /** Filename. */
-  readonly filename?: string
+  readonly filename?: string;
   /** File purpose. */
-  readonly purpose?: string
+  readonly purpose?: string;
   /** Processing status (e.g., `"uploaded"`, `"processed"`). */
-  readonly status?: string
+  readonly status?: string;
 }
 
 /** Purpose of an uploaded file. */
@@ -796,27 +829,27 @@ export declare enum FinishReason {
 /** Function call details. */
 export interface FunctionCall {
   /** Function name. */
-  readonly name: string
+  readonly name: string;
   /** Arguments as a JSON string (parse with serde_json::from_str). */
-  readonly arguments: string
+  readonly arguments: string;
 }
 
 /** Function definition exposed to the model. */
 export interface FunctionDefinition {
   /** Name of the function. Required and must be alphanumeric + underscores. */
-  readonly name: string
+  readonly name: string;
   /** Human-readable description explaining what the function does. */
-  readonly description?: string
+  readonly description?: string;
   /** JSON Schema defining the function's parameters. */
-  readonly parameters?: JsonValue
+  readonly parameters?: JsonValue;
   /** If true, enforce strict JSON schema validation for arguments. */
-  readonly strict?: boolean
+  readonly strict?: boolean;
 }
 
 /** Deprecated legacy function-role message body. */
 export interface FunctionMessage {
-  readonly content?: string
-  readonly name?: string
+  readonly content?: string;
+  readonly name?: string;
 }
 
 /**
@@ -833,7 +866,7 @@ export interface HealthChecker {
    * move it into the returned future without a clone, making the
    * `'static + Send` bound on the future trivially satisfiable.
    */
-  check(upstream: string): Promise<string>
+  check(upstream: string): Promise<string>;
 }
 
 /** The result of a single health probe. */
@@ -847,11 +880,11 @@ export declare enum HealthStatus {
 /** A single generated image, returned as either a URL or base64 data. */
 export interface Image {
   /** Image URL (if response_format was "url"). */
-  readonly url?: string
+  readonly url?: string;
   /** Base64-encoded image data (if response_format was "b64_json"). */
-  readonly b64Json?: string
+  readonly b64Json?: string;
   /** The final prompt used to generate the image (DALL-E 3). */
-  readonly revisedPrompt?: string
+  readonly revisedPrompt?: string;
 }
 
 /** Image detail level controlling token cost and processing. */
@@ -867,63 +900,63 @@ export declare enum ImageDetail {
 /** Response containing generated images. */
 export interface ImagesResponse {
   /** Unix timestamp of image creation. */
-  readonly created?: number
+  readonly created?: number;
   /** List of generated images. */
-  readonly data?: Array<Image>
+  readonly data?: Array<Image>;
 }
 
 /** An image URL reference with optional detail level for processing. */
 export interface ImageUrl {
   /** URL of the image (data URI or HTTP/HTTPS URL). */
-  readonly url?: string
+  readonly url?: string;
   /** Detail level: low (512x512), high (2x2 tiles), or auto (model-selected). */
-  readonly detail?: ImageDetail
+  readonly detail?: ImageDetail;
 }
 
 /** An intent prototype: `(intent_name, prototype_embedding, target_model_id)`. */
 export interface IntentPrototype {
   /** Human-readable name for the intent (used in logs/metrics). */
-  readonly name: string
+  readonly name: string;
   /** Pre-computed embedding vector for this intent. */
-  readonly embedding: Array<number>
+  readonly embedding: Array<number>;
   /** Model to route to when this intent is detected. */
-  readonly model: string
+  readonly model: string;
 }
 
 /** JSON Schema specification for constrained output. */
 export interface JsonSchemaFormat {
   /** Name of the schema (must be unique in the request). */
-  readonly name?: string
+  readonly name?: string;
   /** Description of what the schema represents. */
-  readonly description?: string
+  readonly description?: string;
   /** JSON Schema object defining the output structure. */
-  readonly schema?: JsonValue
+  readonly schema?: JsonValue;
   /** If true, enforce strict schema validation. */
-  readonly strict?: boolean
+  readonly strict?: boolean;
 }
 
 /** A chat message in a conversation. */
 export type Message =
-  | { role: 'system'; 0: SystemMessage }
-  | { role: 'user'; 0: UserMessage }
-  | { role: 'assistant'; 0: AssistantMessage }
-  | { role: 'tool'; 0: ToolMessage }
-  | { role: 'developer'; 0: DeveloperMessage }
-  | { role: 'function'; 0: FunctionMessage }
+  | { role: "system"; 0: SystemMessage }
+  | { role: "user"; 0: UserMessage }
+  | { role: "assistant"; 0: AssistantMessage }
+  | { role: "tool"; 0: ToolMessage }
+  | { role: "developer"; 0: DeveloperMessage }
+  | { role: "function"; 0: FunctionMessage };
 
 /** A model available from the API. */
 export interface ModelObject {
   /** Model ID (e.g., `"gpt-4o"`, `"claude-3-5-sonnet"`). */
-  readonly id?: string
+  readonly id?: string;
   /**
    * Always `"model"` from OpenAI-compatible APIs.  Stored as a plain
    * `String` so non-standard provider values do not break deserialization.
    */
-  readonly object?: string
+  readonly object?: string;
   /** Unix timestamp of model creation (or release date). */
-  readonly created?: number
+  readonly created?: number;
   /** Organization or entity that owns the model. */
-  readonly ownedBy?: string
+  readonly ownedBy?: string;
 }
 
 /** Response listing available models. */
@@ -932,61 +965,61 @@ export interface ModelsListResponse {
    * Always `"list"` from OpenAI-compatible APIs.  Stored as a plain
    * `String` so non-standard provider values do not break deserialization.
    */
-  readonly object?: string
+  readonly object?: string;
   /** List of available models. */
-  readonly data?: Array<ModelObject>
+  readonly data?: Array<ModelObject>;
 }
 
 /** Boolean flags for each moderation category. */
 export interface ModerationCategories {
   /** Sexual content. */
-  readonly sexual?: boolean
+  readonly sexual?: boolean;
   /** Hate speech. */
-  readonly hate?: boolean
+  readonly hate?: boolean;
   /** Harassment. */
-  readonly harassment?: boolean
+  readonly harassment?: boolean;
   /** Self-harm content. */
-  readonly selfHarm?: boolean
+  readonly selfHarm?: boolean;
   /** Sexual content involving minors. */
-  readonly sexualMinors?: boolean
+  readonly sexualMinors?: boolean;
   /** Hate speech that threatens violence. */
-  readonly hateThreatening?: boolean
+  readonly hateThreatening?: boolean;
   /** Graphic violence. */
-  readonly violenceGraphic?: boolean
+  readonly violenceGraphic?: boolean;
   /** Intent to self-harm. */
-  readonly selfHarmIntent?: boolean
+  readonly selfHarmIntent?: boolean;
   /** Instructions for self-harm. */
-  readonly selfHarmInstructions?: boolean
+  readonly selfHarmInstructions?: boolean;
   /** Harassment that threatens violence. */
-  readonly harassmentThreatening?: boolean
+  readonly harassmentThreatening?: boolean;
   /** Non-graphic violence. */
-  readonly violence?: boolean
+  readonly violence?: boolean;
 }
 
 /** Confidence scores for each moderation category. */
 export interface ModerationCategoryScores {
   /** Sexual content score. */
-  readonly sexual?: number
+  readonly sexual?: number;
   /** Hate speech score. */
-  readonly hate?: number
+  readonly hate?: number;
   /** Harassment score. */
-  readonly harassment?: number
+  readonly harassment?: number;
   /** Self-harm content score. */
-  readonly selfHarm?: number
+  readonly selfHarm?: number;
   /** Sexual content involving minors score. */
-  readonly sexualMinors?: number
+  readonly sexualMinors?: number;
   /** Hate speech that threatens violence score. */
-  readonly hateThreatening?: number
+  readonly hateThreatening?: number;
   /** Graphic violence score. */
-  readonly violenceGraphic?: number
+  readonly violenceGraphic?: number;
   /** Intent to self-harm score. */
-  readonly selfHarmIntent?: number
+  readonly selfHarmIntent?: number;
   /** Instructions for self-harm score. */
-  readonly selfHarmInstructions?: number
+  readonly selfHarmInstructions?: number;
   /** Harassment that threatens violence score. */
-  readonly harassmentThreatening?: number
+  readonly harassmentThreatening?: number;
   /** Non-graphic violence score. */
-  readonly violence?: number
+  readonly violence?: number;
 }
 
 /** Input to the moderation endpoint — a single string or multiple strings. */
@@ -1000,84 +1033,84 @@ export declare enum ModerationInput {
 /** Request to classify content for policy violations. */
 export interface ModerationRequest {
   /** Text or texts to check. */
-  readonly input?: ModerationInput
+  readonly input?: ModerationInput;
   /** Model ID (e.g., `"text-moderation-latest"`). Optional; API uses default if unset. */
-  readonly model?: string
+  readonly model?: string;
 }
 
 /** Response from the moderation endpoint. */
 export interface ModerationResponse {
   /** Unique identifier for this moderation request. */
-  readonly id: string
+  readonly id: string;
   /** Model used for classification. */
-  readonly model: string
+  readonly model: string;
   /** Results for each input string. */
-  readonly results: Array<ModerationResult>
+  readonly results: Array<ModerationResult>;
 }
 
 /** A single moderation classification result. */
 export interface ModerationResult {
   /** True if any category was flagged. */
-  readonly flagged: boolean
+  readonly flagged: boolean;
   /** Boolean flags for each moderation category. */
-  readonly categories: ModerationCategories
+  readonly categories: ModerationCategories;
   /** Confidence scores for each category. */
-  readonly categoryScores: ModerationCategoryScores
+  readonly categoryScores: ModerationCategoryScores;
 }
 
 /** Document input for OCR — either a URL or inline base64 data. */
 export type OcrDocument =
-  | { type: 'document_url'; url: string }
-  | { type: 'base64'; data: string; mediaType: string }
+  | { type: "document_url"; url: string }
+  | { type: "base64"; data: string; mediaType: string };
 
 /** An image extracted from an OCR page. */
 export interface OcrImage {
   /** Unique image identifier within the document. */
-  readonly id: string
+  readonly id: string;
   /** Base64-encoded image data (if `include_image_base64` was true). */
-  readonly imageBase64?: string
+  readonly imageBase64?: string;
 }
 
 /** A single page of OCR output. */
 export interface OcrPage {
   /** Page index (0-based). */
-  readonly index: number
+  readonly index: number;
   /** Extracted page content as Markdown. */
-  readonly markdown: string
+  readonly markdown: string;
   /** Embedded images extracted from the page (if `include_image_base64` was true). */
-  readonly images?: Array<OcrImage>
+  readonly images?: Array<OcrImage>;
   /** Page dimensions in pixels, if available. */
-  readonly dimensions?: PageDimensions
+  readonly dimensions?: PageDimensions;
 }
 
 /** An OCR request. */
 export interface OcrRequest {
   /** The model/provider to use (e.g. `"mistral/mistral-ocr-latest"`). */
-  readonly model?: string
+  readonly model?: string;
   /** The document to process (URL or base64). */
-  readonly document?: OcrDocument
+  readonly document?: OcrDocument;
   /** Specific pages to process (1-indexed). `None` means all pages. */
-  readonly pages?: Array<number>
+  readonly pages?: Array<number>;
   /** Whether to include base64-encoded images of each processed page. */
-  readonly includeImageBase64?: boolean
+  readonly includeImageBase64?: boolean;
 }
 
 /** An OCR response. */
 export interface OcrResponse {
   /** Extracted pages in order. */
-  readonly pages: Array<OcrPage>
+  readonly pages: Array<OcrPage>;
   /** Model/provider used for OCR. */
-  readonly model: string
+  readonly model: string;
   /** Token usage, if reported by the provider. */
-  readonly usage?: Usage
+  readonly usage?: Usage;
 }
 
 /** Page dimensions in pixels. */
 export interface PageDimensions {
   /** Width in pixels. */
-  readonly width: number
+  readonly width: number;
   /** Height in pixels. */
-  readonly height: number
+  readonly height: number;
 }
 
 /**
@@ -1090,9 +1123,9 @@ export interface PageDimensions {
  */
 export interface PromptTokensDetails {
   /** Cached tokens present in the prompt. Defaults to 0 when absent. */
-  readonly cachedTokens?: number
+  readonly cachedTokens?: number;
   /** Audio input tokens present in the prompt. Defaults to 0 when absent. */
-  readonly audioTokens?: number
+  readonly audioTokens?: number;
 }
 
 /**
@@ -1122,42 +1155,40 @@ export interface PromptTokensDetails {
  */
 export interface ProviderCapabilities {
   /** The provider accepts image input in chat messages. */
-  readonly vision?: boolean
+  readonly vision?: boolean;
   /** The provider supports extended-thinking / reasoning tokens. */
-  readonly reasoning?: boolean
+  readonly reasoning?: boolean;
   /** The provider supports JSON-mode or `response_format` structured output. */
-  readonly structuredOutput?: boolean
+  readonly structuredOutput?: boolean;
   /** The provider supports tool / function calling. */
-  readonly functionCalling?: boolean
+  readonly functionCalling?: boolean;
   /** The provider accepts audio as input. */
-  readonly audioIn?: boolean
+  readonly audioIn?: boolean;
   /** The provider can generate audio / TTS output. */
-  readonly audioOut?: boolean
+  readonly audioOut?: boolean;
   /** The provider accepts video as input. */
-  readonly videoIn?: boolean
+  readonly videoIn?: boolean;
 }
 
 /**
  * Static configuration for a single provider entry in providers.json.
  *
  * This struct deliberately does not include capability flags or streaming
- * format, which are accessed via the [`capabilities`] function.  Keeping
- * these fields separate preserves backward compatibility with all generated
- * binding code that constructs `ProviderConfig` using struct literal syntax.
+ * format, which are accessed via the [`capabilities`] function.
  */
 export interface ProviderConfig {
   /** Provider identifier (matches the entry key in providers.json). */
-  readonly name: string
+  readonly name: string;
   /** Human-readable provider name shown in UIs. */
-  readonly displayName?: string
+  readonly displayName?: string;
   /** Base URL used as the default for this provider's HTTP client. */
-  readonly baseUrl?: string
+  readonly baseUrl?: string;
   /** Authentication scheme metadata (auth type + env var holding the key). */
-  readonly auth?: AuthConfig
+  readonly auth?: AuthConfig;
   /** Supported endpoint kinds (e.g. `chat`, `embeddings`). */
-  readonly endpoints?: Array<string>
+  readonly endpoints?: Array<string>;
   /** Model-name prefixes claimed by this provider (e.g. `["gpt-", "o1-"]`). */
-  readonly modelPrefixes?: Array<string>
+  readonly modelPrefixes?: Array<string>;
   /**
    * Parameter key renaming for this provider.
    *
@@ -1165,17 +1196,17 @@ export interface ProviderConfig {
    * to the name this provider expects (e.g. `"max_tokens"`).  Applied
    * automatically by [`ConfigDrivenProvider::transform_request`].
    */
-  readonly paramMappings?: Record<string, string>
+  readonly paramMappings?: Record<string, string>;
 }
 
 /** Configuration for per-model rate limits. */
 export interface RateLimitConfig {
   /** Maximum requests per window.  `None` means unlimited. */
-  readonly rpm?: number
+  readonly rpm?: number;
   /** Maximum tokens per window.  `None` means unlimited. */
-  readonly tpm?: number
+  readonly tpm?: number;
   /** Fixed window duration (defaults to 60 s). */
-  readonly window?: number
+  readonly window?: number;
 }
 
 /** Controls how much reasoning effort the model should use. */
@@ -1196,151 +1227,149 @@ export declare enum RerankDocument {
 /** Request to rerank documents by relevance to a query. */
 export interface RerankRequest {
   /** Model ID (e.g., `"cohere/rerank-english-v3.0"`). */
-  readonly model?: string
+  readonly model?: string;
   /** The search query. */
-  readonly query?: string
+  readonly query?: string;
   /** Documents to rerank. */
-  readonly documents?: Array<RerankDocument>
+  readonly documents?: Array<RerankDocument>;
   /** Return only the top N results. Optional. */
-  readonly topN?: number
+  readonly topN?: number;
   /** Include the document content in results. Defaults to false. */
-  readonly returnDocuments?: boolean
+  readonly returnDocuments?: boolean;
 }
 
 /** Response from the rerank endpoint. */
 export interface RerankResponse {
   /** Unique identifier for this rerank request. */
-  readonly id?: string
+  readonly id?: string;
   /** Reranked documents in order of relevance. */
-  readonly results: Array<RerankResult>
+  readonly results: Array<RerankResult>;
   /** Optional metadata about the reranking operation. */
-  readonly meta?: JsonValue
+  readonly meta?: JsonValue;
 }
 
 /** A single reranked document with its relevance score. */
 export interface RerankResult {
   /** Original document index in the input list. */
-  readonly index: number
+  readonly index: number;
   /** Relevance score in `[0, 1]`. Higher indicates more relevant. */
-  readonly relevanceScore: number
+  readonly relevanceScore: number;
   /** Original document content (if `return_documents` was true). */
-  readonly document?: RerankResultDocument
+  readonly document?: RerankResultDocument;
 }
 
 /** The text content of a reranked document, returned when `return_documents` is true. */
 export interface RerankResultDocument {
   /** Document text. */
-  readonly text: string
+  readonly text: string;
 }
 
 /** Response format constraint. */
 export type ResponseFormat =
-  | { type: 'text' }
-  | { type: 'json_object' }
-  | { type: 'json_schema'; jsonSchema: JsonSchemaFormat }
+  | { type: "text" }
+  | { type: "json_object" }
+  | { type: "json_schema"; jsonSchema: JsonSchemaFormat };
 
 /** Response from a structured response request. */
 export interface ResponseObject {
   /** Unique response ID. */
-  readonly id?: string
+  readonly id?: string;
   /** Object type (e.g., `"response"`). */
-  readonly object?: string
+  readonly object?: string;
   /** Unix timestamp of response creation. */
-  readonly createdAt?: number
+  readonly createdAt?: number;
   /** Model used to generate the response. */
-  readonly model?: string
+  readonly model?: string;
   /** Status (e.g., `"succeeded"`, `"failed"`). */
-  readonly status?: string
+  readonly status?: string;
   /** Output items from the response. */
-  readonly output?: Array<ResponseOutputItem>
+  readonly output?: Array<ResponseOutputItem>;
   /** Token usage. */
-  readonly usage?: ResponseUsage
+  readonly usage?: ResponseUsage;
   /** Error details (if status is "failed"). */
-  readonly error?: JsonValue
+  readonly error?: JsonValue;
 }
 
 /** A single output item from the response. */
 export interface ResponseOutputItem {
   /** Output type (e.g., `"text"`, `"object"`, `"error"`). */
-  readonly itemType?: string
+  readonly itemType?: string;
   /** Output content (flattened into the object). */
-  readonly content?: JsonValue
+  readonly content?: JsonValue;
 }
 
 /** A tool available for the response request. */
 export interface ResponseTool {
   /** Tool type (e.g., "extractor", "search"). */
-  readonly toolType?: string
+  readonly toolType?: string;
   /** Tool configuration (flattened into the object). */
-  readonly config?: JsonValue
+  readonly config?: JsonValue;
 }
 
 /** Token usage for a response. */
 export interface ResponseUsage {
   /** Input tokens used. */
-  readonly inputTokens?: number
+  readonly inputTokens?: number;
   /** Output tokens used. */
-  readonly outputTokens?: number
+  readonly outputTokens?: number;
   /** Total tokens used. */
-  readonly totalTokens?: number
+  readonly totalTokens?: number;
 }
 
 /** A search request. */
 export interface SearchRequest {
   /** The model/provider to use (e.g. `"brave/web-search"`, `"tavily/search"`). */
-  readonly model?: string
+  readonly model?: string;
   /** The search query string. */
-  readonly query?: string
+  readonly query?: string;
   /** Maximum number of results to return. */
-  readonly maxResults?: number
+  readonly maxResults?: number;
   /** Domain filter — restrict results to specific domains. */
-  readonly searchDomainFilter?: Array<string>
+  readonly searchDomainFilter?: Array<string>;
   /** Country code for localized results (ISO 3166-1 alpha-2, e.g., `"US"`, `"FR"`). */
-  readonly country?: string
+  readonly country?: string;
 }
 
 /** A search response. */
 export interface SearchResponse {
   /** List of search results. */
-  readonly results: Array<SearchResult>
+  readonly results: Array<SearchResult>;
   /** Model/provider that performed the search. */
-  readonly model: string
+  readonly model: string;
 }
 
 /** An individual search result. */
 export interface SearchResult {
   /** Result title. */
-  readonly title: string
+  readonly title: string;
   /** Result URL. */
-  readonly url: string
+  readonly url: string;
   /** Text snippet or excerpt from the page. */
-  readonly snippet: string
+  readonly snippet: string;
   /** Publication or last-updated date, if available. */
-  readonly date?: string
+  readonly date?: string;
 }
 
 /**
  * The value broadcast from a singleflight leader to all followers.
  *
- * `Arc<LiterLlmError>` is used because `LiterLlmError` is not `Clone` and
- * broadcast channels require `T: Clone`.  The `Arc` adds only a reference-count
- * bump per follower, which is negligible under the burst loads this layer targets.
+ * The error value is shared so every follower receives the same upstream
+ * failure without cloning the underlying error.
  */
-export declare class SingleflightResult {
-}
+export declare class SingleflightResult {}
 
 /** Name of the specific function to invoke. */
 export interface SpecificFunction {
   /** Function name. */
-  readonly name?: string
+  readonly name?: string;
 }
 
 /** Directive to call a specific tool. */
 export interface SpecificToolChoice {
   /** Tool type (always "function"). */
-  readonly choiceType?: ToolType
+  readonly choiceType?: ToolType;
   /** The specific function to invoke. */
-  readonly function?: SpecificFunction
+  readonly function?: SpecificFunction;
 }
 
 /** Stop sequence(s) that cause the model to stop generating. */
@@ -1354,25 +1383,25 @@ export declare enum StopSequence {
 /** A streaming choice with incremental delta. */
 export interface StreamChoice {
   /** Index of this choice in the choices array. */
-  readonly index?: number
+  readonly index?: number;
   /** Incremental update to the message (content, tool calls, etc.). */
-  readonly delta?: StreamDelta
+  readonly delta?: StreamDelta;
   /** Why the stream ended (present only in final chunk). */
-  readonly finishReason?: FinishReason
+  readonly finishReason?: FinishReason;
 }
 
 /** Incremental delta in a stream chunk. */
 export interface StreamDelta {
   /** Role (typically present only in the first chunk). */
-  readonly role?: string
+  readonly role?: string;
   /** Partial content chunk (e.g., a few words of the response). */
-  readonly content?: string
+  readonly content?: string;
   /** Partial tool calls being streamed. */
-  readonly toolCalls?: Array<StreamToolCall>
+  readonly toolCalls?: Array<StreamToolCall>;
   /** Deprecated legacy function_call delta; retained for API compatibility. */
-  readonly functionCall?: StreamFunctionCall
+  readonly functionCall?: StreamFunctionCall;
   /** Partial refusal message. */
-  readonly refusal?: string
+  readonly refusal?: string;
 }
 
 /**
@@ -1393,45 +1422,45 @@ export declare enum StreamFormat {
 /** Partial function call details in a stream. */
 export interface StreamFunctionCall {
   /** Function name (typically in the first chunk). */
-  readonly name?: string
+  readonly name?: string;
   /** Partial JSON arguments chunk. */
-  readonly arguments?: string
+  readonly arguments?: string;
 }
 
 /** Options for streaming responses. */
 export interface StreamOptions {
   /** If true, include token usage in the final stream chunk. */
-  readonly includeUsage?: boolean
+  readonly includeUsage?: boolean;
 }
 
 /** A streaming tool call being built incrementally. */
 export interface StreamToolCall {
   /** Index of this tool call in the tool_calls array. */
-  readonly index?: number
+  readonly index?: number;
   /** Tool call ID (typically in the first chunk for this call). */
-  readonly id?: string
+  readonly id?: string;
   /** Tool type (typically "function"). */
-  readonly callType?: ToolType
+  readonly callType?: ToolType;
   /** Partial function name and arguments. */
-  readonly function?: StreamFunctionCall
+  readonly function?: StreamFunctionCall;
 }
 
 /** System message guiding model behavior for the entire conversation. */
 export interface SystemMessage {
   /** Instructions or context that apply throughout the conversation. */
-  readonly content?: string
+  readonly content?: string;
   /** Optional name for the system message source. */
-  readonly name?: string
+  readonly name?: string;
 }
 
 /** A tool call the model wants to execute. */
 export interface ToolCall {
   /** Unique ID for this call, used to reference in tool result messages. */
-  readonly id: string
+  readonly id: string;
   /** Tool type (always "function"). */
-  readonly callType: ToolType
+  readonly callType: ToolType;
   /** Function name and arguments. */
-  readonly function: FunctionCall
+  readonly function: FunctionCall;
 }
 
 /** Tool usage mode or a specific tool to call. */
@@ -1455,11 +1484,11 @@ export declare enum ToolChoiceMode {
 /** Tool execution result returned to the model. */
 export interface ToolMessage {
   /** Result of the tool execution. */
-  readonly content?: string
+  readonly content?: string;
   /** ID of the tool call this result responds to. */
-  readonly toolCallId?: string
+  readonly toolCallId?: string;
   /** Optional tool/function name. */
-  readonly name?: string
+  readonly name?: string;
 }
 
 /**
@@ -1476,41 +1505,41 @@ export declare enum ToolType {
 /** Response from a transcription request. */
 export interface TranscriptionResponse {
   /** The transcribed text. */
-  readonly text?: string
+  readonly text?: string;
   /** Detected language (ISO-639-1 code). */
-  readonly language?: string
+  readonly language?: string;
   /** Total audio duration in seconds. */
-  readonly duration?: number
+  readonly duration?: number;
   /** Detailed segment-level transcription (if response_format is "verbose_json"). */
-  readonly segments?: Array<TranscriptionSegment>
+  readonly segments?: Array<TranscriptionSegment>;
 }
 
 /** A segment of transcribed audio with timing information. */
 export interface TranscriptionSegment {
   /** Segment index (0-based). */
-  readonly id?: number
+  readonly id?: number;
   /** Start time in seconds. */
-  readonly start?: number
+  readonly start?: number;
   /** End time in seconds. */
-  readonly end?: number
+  readonly end?: number;
   /** Transcribed text for this segment. */
-  readonly text?: string
+  readonly text?: string;
 }
 
 /** Token-usage accounting returned by the provider on each completion / embedding call. */
 export interface Usage {
   /** Prompt tokens used. Defaults to 0 when absent (some providers omit this). */
-  readonly promptTokens?: number
+  readonly promptTokens?: number;
   /** Completion tokens used. Defaults to 0 when absent (e.g. embedding responses). */
-  readonly completionTokens?: number
+  readonly completionTokens?: number;
   /** Total tokens used. Defaults to 0 when absent (some providers omit this). */
-  readonly totalTokens?: number
+  readonly totalTokens?: number;
   /**
    * Breakdown of tokens used in the prompt, including cached tokens served
    * at the provider's discounted cache-read rate. Absent when the provider
    * does not return prompt-token details.
    */
-  readonly promptTokensDetails?: PromptTokensDetails
+  readonly promptTokensDetails?: PromptTokensDetails;
 }
 
 /** User message content as either plain text or a list of multimodal parts. */
@@ -1524,9 +1553,9 @@ export declare enum UserContent {
 /** User message in the conversation. */
 export interface UserMessage {
   /** Message content as plain text or array of content parts (text, images, documents, audio). */
-  readonly content?: UserContent
+  readonly content?: UserContent;
   /** Optional name for the user. */
-  readonly name?: string
+  readonly name?: string;
 }
 
 /**
@@ -1537,13 +1566,13 @@ export interface UserMessage {
  */
 export interface WaitForBatchConfig {
   /** Initial interval between polls, in seconds. */
-  readonly initialIntervalSecs?: number
+  readonly initialIntervalSecs?: number;
   /** Maximum interval between polls (backoff plateau), in seconds. */
-  readonly maxIntervalSecs?: number
+  readonly maxIntervalSecs?: number;
   /** Exponential backoff multiplier (e.g., 1.5 increases delay by 50% each poll). */
-  readonly backoffMultiplier?: number
+  readonly backoffMultiplier?: number;
   /** Optional timeout in seconds — polling fails if this duration is exceeded. */
-  readonly timeoutSecs?: number
+  readonly timeoutSecs?: number;
 }
 
 /**
@@ -1561,17 +1590,17 @@ export declare function registerCustomProvider(config: CustomProviderConfig): vo
  *
  * Returns `true` if a provider with the given name was found and removed,
  * `false` if no such provider existed.
- * @throws Returns an error only if the internal lock is poisoned.
+ * @throws Returns an error if the custom-provider registry cannot be updated.
  */
 export declare function unregisterCustomProvider(name: string): boolean;
 
 export declare class ChatStreamIterator {
-  next(value?: undefined): Promise<IteratorResult<ChatCompletionChunk, void>>
-  [Symbol.asyncIterator](): AsyncGenerator<ChatCompletionChunk, void, undefined>
+  next(value?: undefined): Promise<IteratorResult<ChatCompletionChunk, void>>;
+  [Symbol.asyncIterator](): AsyncGenerator<ChatCompletionChunk, void, undefined>;
 }
 
 export declare class LiterLlmErrorInfo {
-  statusCode(): number
-  isTransient(): boolean
-  errorType(): string
+  statusCode(): number;
+  isTransient(): boolean;
+  errorType(): string;
 }