@maximai/maxim-js/src/lib/logger/vercel/types.d.ts

Maxim AI JS SDK. Visit https://getmaxim.ai for more info.

import type { LanguageModelV2CallWarning, LanguageModelV2Content, LanguageModelV2FinishReason, LanguageModelV2ResponseMetadata, LanguageModelV2Usage, SharedV2Headers, SharedV2ProviderMetadata } from "ai-sdk-provider-v2";
/**
 * Represents the expected structure of a result from a language model generation call.
 *
 * This interface defines the minimal fields required for converting a generation result into a standardized chat completion result, including token usage and model information.
 *
 * @property usage - Token usage statistics for the generation.
 * @property usage.promptTokens - Number of tokens in the prompt.
 * @property usage.completionTokens - Number of tokens in the completion.
 * @property response - Optional response metadata, including model identifiers.
 * @property response.model_id - The model identifier (snake_case).
 * @property response.modelId - The model identifier (camelCase).
 * @property rawResponse - The raw response object from the model provider.
 */
export interface DoGenerateResultLike {
    usage: {
        promptTokens: number;
        completionTokens: number;
    };
    response?: {
        model_id?: string;
        modelId?: string;
    };
    rawResponse?: any;
}
export type DoGenerateV2Result = {
    content: Array<LanguageModelV2Content>;
    finishReason: LanguageModelV2FinishReason;
    usage: LanguageModelV2Usage;
    providerMetadata?: SharedV2ProviderMetadata;
    request?: {
        body?: unknown;
    };
    response?: LanguageModelV2ResponseMetadata & {
        headers?: SharedV2Headers;
        body?: unknown;
    };
    warnings: Array<LanguageModelV2CallWarning>;
};
export type LanguageFirstTokenModel = {
    received: boolean;
    time: number | null;
};
/**
 * Metadata options for Maxim tracing integration with Vercel AI SDK providers.
 *
 * This type allows you to attach custom metadata to sessions, traces, generations, and spans when using Maxim's tracing/logging features. These fields enable advanced tracking, naming, and tagging of AI model calls for observability and debugging.
 *
 * @property sessionId - Link your traces to a session by specifying its ID.
 * @property sessionName - Override the default session name for this trace.
 * @property sessionTags - Add custom tags to the session for filtering or grouping.
 * @property traceId - Pass in an existing trace's ID to associate this call with a specific trace.
 * @property traceName - Override the default trace name for this call.
 * @property traceTags - Add custom tags to the trace for filtering or grouping.
 * @property generationName - Provide a custom name for the generation (model output) event.
 * @property generationTags - Add custom tags to the generation for filtering or grouping.
 * @property spanId - Pass in a specific span ID to link this call to a particular span.
 * @property spanName - Override the default span name for this call.
 * @property spanTags - Add custom tags to the span for filtering or grouping.
 */
export type MaximVercelProviderMetadata = {
    /** Link your traces to a session */
    sessionId?: string;
    /** Override session name */
    sessionName?: string;
    /** Add tags to session */
    sessionTags?: Record<string, string>;
    /** Pass in an existing trace's id */
    traceId?: string;
    /** Override trace name */
    traceName?: string;
    /** Add tags to trace */
    traceTags?: Record<string, string>;
    /** Pass in a custom generation name */
    generationName?: string;
    /** Add tags to generation */
    generationTags?: Record<string, string>;
    /** Pass in a specific span id */
    spanId?: string;
    /** Override span name */
    spanName?: string;
    /** Add tags to span */
    spanTags?: Record<string, string>;
};