@tanstack/ai 0.36.0 → 0.37.0

package/dist/esm/activities/generateAudio/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sources":["../../../../src/activities/generateAudio/index.ts"],"sourcesContent":["/**\n * Audio Generation Activity\n *\n * Generates audio (music, sound effects, etc.) from text prompts.\n * This is a self-contained module with implementation, types, and JSDoc.\n */\n\nimport { aiEventClient } from '@tanstack/ai-event-client'\nimport { streamGenerationResult } from '../stream-generation-result.js'\nimport { resolveDebugOption } from '../../logger/resolve'\nimport {\n  createGenerationContext,\n  runGenerationError,\n  runGenerationFinish,\n  runGenerationStart,\n  runGenerationUsage,\n} from '../middleware'\nimport type { InternalLogger } from '../../logger/internal-logger'\nimport type { DebugOption } from '../../logger/types'\nimport type { GenerationMiddleware } from '../middleware'\nimport type { AudioAdapter } from './adapter'\nimport type { AudioGenerationResult, StreamChunk } from '../../types'\n\n// ===========================\n// Activity Kind\n// ===========================\n\n/** The adapter kind this activity handles */\nexport const kind = 'audio' as const\n\n// ===========================\n// Type Extraction Helpers\n// ===========================\n\n/**\n * Extract provider options from an AudioAdapter via ~types.\n */\nexport type AudioProviderOptions<TAdapter> = TAdapter extends {\n  '~types': { providerOptions: infer P extends object }\n}\n  ? P\n  : object\n\n// ===========================\n// Activity Options Type\n// ===========================\n\n/**\n * Options for the audio generation activity.\n * The model is extracted from the adapter's model property.\n *\n * @template TAdapter - The audio adapter type\n * @template TStream - Whether to stream the output\n */\nexport interface AudioActivityOptions<\n  TAdapter extends AudioAdapter<string, AudioProviderOptions<TAdapter>>,\n  TStream extends boolean = false,\n> {\n  /** The audio adapter to use (must be created with a model) */\n  adapter: TAdapter & { kind: typeof kind }\n  /** Text description of the desired audio */\n  prompt: string\n  /** Desired duration in seconds */\n  duration?: number\n  /** Provider-specific options for audio generation */\n  modelOptions?: AudioProviderOptions<TAdapter>\n  /**\n   * Whether to stream the generation result.\n   * When true, returns an AsyncIterable<StreamChunk> for streaming transport.\n   * When false or not provided, returns a Promise<AudioGenerationResult>.\n   *\n   * @default false\n   */\n  stream?: TStream\n  /**\n   * Enable debug logging. Pass `true` to enable all categories, `false` to\n   * silence everything including errors, or a `DebugConfig` object for granular\n   * control and/or a custom `Logger`.\n   */\n  debug?: DebugOption\n  /**\n   * Observe-only middleware notified on start, usage, success, and error. Pass\n   * `otelMiddleware()` to emit OpenTelemetry spans, or implement the\n   * `GenerationMiddleware` contract for a custom backend.\n   */\n  middleware?: Array<GenerationMiddleware>\n}\n\n// ===========================\n// Activity Result Type\n// ===========================\n\n/**\n * Result type for the audio generation activity.\n * - If stream is true: AsyncIterable<StreamChunk>\n * - Otherwise: Promise<AudioGenerationResult>\n */\nexport type AudioActivityResult<TStream extends boolean = false> =\n  TStream extends true\n    ? AsyncIterable<StreamChunk>\n    : Promise<AudioGenerationResult>\n\nfunction createId(prefix: string): string {\n  return `${prefix}-${Date.now()}-${Math.random().toString(36).slice(2, 9)}`\n}\n\n// ===========================\n// Activity Implementation\n// ===========================\n\n/**\n * Audio generation activity - generates audio from text prompts.\n *\n * Uses AI models to create music, sound effects, and other audio content.\n *\n * @example Generate music from a prompt\n * ```ts\n * import { generateAudio } from '@tanstack/ai'\n * import { falAudio } from '@tanstack/ai-fal'\n *\n * const result = await generateAudio({\n *   adapter: falAudio('fal-ai/diffrhythm'),\n *   prompt: 'An upbeat electronic track with synths',\n *   duration: 10\n * })\n *\n * console.log(result.audio.url) // URL to generated audio\n * ```\n */\nexport function generateAudio<\n  TAdapter extends AudioAdapter<string, AudioProviderOptions<TAdapter>>,\n  TStream extends boolean = false,\n>(\n  options: AudioActivityOptions<TAdapter, TStream>,\n): AudioActivityResult<TStream> {\n  if (options.stream) {\n    return streamGenerationResult(() =>\n      runGenerateAudio(options),\n    ) as AudioActivityResult<TStream>\n  }\n  return runGenerateAudio(options) as AudioActivityResult<TStream>\n}\n\n/**\n * Run the core audio generation logic (non-streaming).\n */\nasync function runGenerateAudio<\n  TAdapter extends AudioAdapter<string, AudioProviderOptions<TAdapter>>,\n>(\n  options: AudioActivityOptions<TAdapter, boolean>,\n): Promise<AudioGenerationResult> {\n  const {\n    adapter,\n    stream: _stream,\n    debug: _debug,\n    middleware,\n    ...rest\n  } = options\n  const model = adapter.model\n  const requestId = createId('audio')\n  const startTime = Date.now()\n  const logger: InternalLogger = resolveDebugOption(options.debug)\n  const providerName =\n    (adapter as { name?: string; provider?: string }).provider ??\n    (adapter as { name?: string }).name ??\n    'unknown'\n\n  const mwCtx = createGenerationContext({\n    requestId,\n    activity: 'audio',\n    provider: adapter.name,\n    model,\n    modelOptions: rest.modelOptions,\n    createId,\n  })\n\n  await runGenerationStart(middleware, mwCtx)\n\n  aiEventClient.emit('audio:request:started', {\n    requestId,\n    provider: adapter.name,\n    model,\n    prompt: rest.prompt,\n    duration: rest.duration,\n    modelOptions: rest.modelOptions as Record<string, unknown> | undefined,\n    timestamp: startTime,\n  })\n\n  logger.request(`activity=generateAudio provider=${providerName}`, {\n    provider: providerName,\n    model,\n  })\n\n  try {\n    const result = await adapter.generateAudio({ ...rest, model, logger })\n    const elapsedMs = Date.now() - startTime\n\n    aiEventClient.emit('audio:request:completed', {\n      requestId,\n      provider: adapter.name,\n      model,\n      audio: result.audio,\n      duration: elapsedMs,\n      modelOptions: rest.modelOptions as Record<string, unknown> | undefined,\n      timestamp: Date.now(),\n    })\n\n    if (result.usage) {\n      aiEventClient.emit('audio:usage', {\n        requestId,\n        model,\n        usage: result.usage,\n        modelOptions: rest.modelOptions as Record<string, unknown> | undefined,\n        timestamp: Date.now(),\n      })\n    }\n\n    logger.output(`activity=generateAudio provider=${providerName}`, {\n      contentType: result.audio.contentType,\n      audioDuration: result.audio.duration,\n    })\n\n    if (result.usage) await runGenerationUsage(middleware, mwCtx, result.usage)\n    await runGenerationFinish(middleware, mwCtx, {\n      duration: elapsedMs,\n      usage: result.usage,\n    })\n\n    return result\n  } catch (error) {\n    const elapsedMs = Date.now() - startTime\n    const err = error as Error\n    aiEventClient.emit('audio:request:error', {\n      requestId,\n      provider: adapter.name,\n      model,\n      error: { message: err.message, name: err.name },\n      duration: elapsedMs,\n      modelOptions: rest.modelOptions as Record<string, unknown> | undefined,\n      timestamp: Date.now(),\n    })\n    await runGenerationError(middleware, mwCtx, {\n      error,\n      duration: elapsedMs,\n    })\n    logger.errors('generateAudio activity failed', {\n      error,\n      source: 'generateAudio',\n    })\n    throw error\n  }\n}\n\n// ===========================\n// Options Factory\n// ===========================\n\n/**\n * Create typed options for the generateAudio() function without executing.\n */\nexport function createAudioOptions<\n  TAdapter extends AudioAdapter<string, AudioProviderOptions<TAdapter>>,\n  TStream extends boolean = false,\n>(\n  options: AudioActivityOptions<TAdapter, TStream>,\n): AudioActivityOptions<TAdapter, TStream> {\n  return options\n}\n\n// Re-export adapter types\nexport type {\n  AudioAdapter,\n  AudioAdapterConfig,\n  AnyAudioAdapter,\n} from './adapter'\nexport { BaseAudioAdapter } from './adapter'\n"],"names":[],"mappings":";;;;AA4BO,MAAM,OAAO;AA0EpB,SAAS,SAAS,QAAwB;AACxC,SAAO,GAAG,MAAM,IAAI,KAAK,IAAA,CAAK,IAAI,KAAK,OAAA,EAAS,SAAS,EAAE,EAAE,MAAM,GAAG,CAAC,CAAC;AAC1E;AAyBO,SAAS,cAId,SAC8B;AAC9B,MAAI,QAAQ,QAAQ;AAClB,WAAO;AAAA,MAAuB,MAC5B,iBAAiB,OAAO;AAAA,IAAA;AAAA,EAE5B;AACA,SAAO,iBAAiB,OAAO;AACjC;AAKA,eAAe,iBAGb,SACgC;AAChC,QAAM;AAAA,IACJ;AAAA,IACA,QAAQ;AAAA,IACR,OAAO;AAAA,IACP;AAAA,IACA,GAAG;AAAA,EAAA,IACD;AACJ,QAAM,QAAQ,QAAQ;AACtB,QAAM,YAAY,SAAS,OAAO;AAClC,QAAM,YAAY,KAAK,IAAA;AACvB,QAAM,SAAyB,mBAAmB,QAAQ,KAAK;AAC/D,QAAM,eACH,QAAiD,YACjD,QAA8B,QAC/B;AAEF,QAAM,QAAQ,wBAAwB;AAAA,IACpC;AAAA,IACA,UAAU;AAAA,IACV,UAAU,QAAQ;AAAA,IAClB;AAAA,IACA,cAAc,KAAK;AAAA,IACnB;AAAA,EAAA,CACD;AAED,QAAM,mBAAmB,YAAY,KAAK;AAE1C,gBAAc,KAAK,yBAAyB;AAAA,IAC1C;AAAA,IACA,UAAU,QAAQ;AAAA,IAClB;AAAA,IACA,QAAQ,KAAK;AAAA,IACb,UAAU,KAAK;AAAA,IACf,cAAc,KAAK;AAAA,IACnB,WAAW;AAAA,EAAA,CACZ;AAED,SAAO,QAAQ,mCAAmC,YAAY,IAAI;AAAA,IAChE,UAAU;AAAA,IACV;AAAA,EAAA,CACD;AAED,MAAI;AACF,UAAM,SAAS,MAAM,QAAQ,cAAc,EAAE,GAAG,MAAM,OAAO,QAAQ;AACrE,UAAM,YAAY,KAAK,IAAA,IAAQ;AAE/B,kBAAc,KAAK,2BAA2B;AAAA,MAC5C;AAAA,MACA,UAAU,QAAQ;AAAA,MAClB;AAAA,MACA,OAAO,OAAO;AAAA,MACd,UAAU;AAAA,MACV,cAAc,KAAK;AAAA,MACnB,WAAW,KAAK,IAAA;AAAA,IAAI,CACrB;AAED,QAAI,OAAO,OAAO;AAChB,oBAAc,KAAK,eAAe;AAAA,QAChC;AAAA,QACA;AAAA,QACA,OAAO,OAAO;AAAA,QACd,cAAc,KAAK;AAAA,QACnB,WAAW,KAAK,IAAA;AAAA,MAAI,CACrB;AAAA,IACH;AAEA,WAAO,OAAO,mCAAmC,YAAY,IAAI;AAAA,MAC/D,aAAa,OAAO,MAAM;AAAA,MAC1B,eAAe,OAAO,MAAM;AAAA,IAAA,CAC7B;AAED,QAAI,OAAO,MAAO,OAAM,mBAAmB,YAAY,OAAO,OAAO,KAAK;AAC1E,UAAM,oBAAoB,YAAY,OAAO;AAAA,MAC3C,UAAU;AAAA,MACV,OAAO,OAAO;AAAA,IAAA,CACf;AAED,WAAO;AAAA,EACT,SAAS,OAAO;AACd,UAAM,YAAY,KAAK,IAAA,IAAQ;AAC/B,UAAM,MAAM;AACZ,kBAAc,KAAK,uBAAuB;AAAA,MACxC;AAAA,MACA,UAAU,QAAQ;AAAA,MAClB;AAAA,MACA,OAAO,EAAE,SAAS,IAAI,SAAS,MAAM,IAAI,KAAA;AAAA,MACzC,UAAU;AAAA,MACV,cAAc,KAAK;AAAA,MACnB,WAAW,KAAK,IAAA;AAAA,IAAI,CACrB;AACD,UAAM,mBAAmB,YAAY,OAAO;AAAA,MAC1C;AAAA,MACA,UAAU;AAAA,IAAA,CACX;AACD,WAAO,OAAO,iCAAiC;AAAA,MAC7C;AAAA,MACA,QAAQ;AAAA,IAAA,CACT;AACD,UAAM;AAAA,EACR;AACF;AASO,SAAS,mBAId,SACyC;AACzC,SAAO;AACT;"}

package/dist/esm/activities/generateImage/adapter.d.ts

@@ -0,0 +1,78 @@
+import { ImageGenerationOptions, ImageGenerationResult, ModelInputModalitiesByName } from '../../types.js';
+/**
+ * Resolve the size type for a model from the model-size map.
+ * If the map has an index signature (i.e. no explicit keys), falls back to string.
+ * If the model is an explicit key, uses its mapped size type.
+ * Otherwise falls back to string.
+ */
+/**
+ * Configuration for image adapter instances
+ */
+export interface ImageAdapterConfig {
+    apiKey?: string;
+    baseUrl?: string;
+    timeout?: number;
+    maxRetries?: number;
+    headers?: Record<string, string>;
+}
+/**
+ * Image adapter interface with pre-resolved generics.
+ *
+ * An adapter is created by a provider function: `provider('model')` → `adapter`
+ * All type resolution happens at the provider call site, not in this interface.
+ *
+ * Generic parameters:
+ * - TModel: The specific model name (e.g., 'dall-e-3')
+ * - TProviderOptions: Base provider-specific options (already resolved)
+ * - TModelProviderOptionsByName: Map from model name to its specific provider options
+ * - TModelSizeByName: Map from model name to its supported sizes
+ * - TModelInputModalitiesByName: Map from model name to the non-text prompt
+ *   modalities it accepts (constrains the `prompt` part types at compile time)
+ */
+export interface ImageAdapter<TModel extends string = string, TProviderOptions extends object = Record<string, unknown>, TModelProviderOptionsByName extends Record<string, any> = Record<string, any>, TModelSizeByName extends Record<string, string | undefined> = Record<string, string>, TModelInputModalitiesByName extends ModelInputModalitiesByName = ModelInputModalitiesByName> {
+    /** Discriminator for adapter kind - used by generate() to determine API shape */
+    readonly kind: 'image';
+    /** Adapter name identifier */
+    readonly name: string;
+    /** The model this adapter is configured for */
+    readonly model: TModel;
+    /**
+     * @internal Type-only properties for inference. Not assigned at runtime.
+     */
+    '~types': {
+        providerOptions: TProviderOptions;
+        modelProviderOptionsByName: TModelProviderOptionsByName;
+        modelSizeByName: TModelSizeByName;
+        modelInputModalitiesByName: TModelInputModalitiesByName;
+    };
+    /**
+     * Generate images from a prompt
+     */
+    generateImages: (options: ImageGenerationOptions<TProviderOptions, TModelSizeByName[TModel]>) => Promise<ImageGenerationResult>;
+}
+/**
+ * An ImageAdapter with any/unknown type parameters.
+ * Useful as a constraint in generic functions and interfaces.
+ */
+export type AnyImageAdapter = ImageAdapter<any, any, any, any, any>;
+/**
+ * Abstract base class for image generation adapters.
+ * Extend this class to implement an image adapter for a specific provider.
+ *
+ * Generic parameters match ImageAdapter - all pre-resolved by the provider function.
+ */
+export declare abstract class BaseImageAdapter<TModel extends string = string, TProviderOptions extends object = Record<string, unknown>, TModelProviderOptionsByName extends Record<string, any> = Record<string, any>, TModelSizeByName extends Record<string, string | undefined> = Record<string, string>, TModelInputModalitiesByName extends ModelInputModalitiesByName = ModelInputModalitiesByName> implements ImageAdapter<TModel, TProviderOptions, TModelProviderOptionsByName, TModelSizeByName, TModelInputModalitiesByName> {
+    readonly kind: "image";
+    abstract readonly name: string;
+    readonly model: TModel;
+    '~types': {
+        providerOptions: TProviderOptions;
+        modelProviderOptionsByName: TModelProviderOptionsByName;
+        modelSizeByName: TModelSizeByName;
+        modelInputModalitiesByName: TModelInputModalitiesByName;
+    };
+    protected config: ImageAdapterConfig;
+    constructor(model: TModel, config?: ImageAdapterConfig);
+    abstract generateImages(options: ImageGenerationOptions<TProviderOptions, TModelSizeByName[TModel]>): Promise<ImageGenerationResult>;
+    protected generateId(): string;
+}

package/dist/esm/activities/generateImage/adapter.js

@@ -0,0 +1,16 @@
+class BaseImageAdapter {
+  kind = "image";
+  model;
+  config;
+  constructor(model, config = {}) {
+    this.config = config;
+    this.model = model;
+  }
+  generateId() {
+    return `${this.name}-${Date.now()}-${Math.random().toString(36).substring(7)}`;
+  }
+}
+export {
+  BaseImageAdapter
+};
+//# sourceMappingURL=adapter.js.map

package/dist/esm/activities/generateImage/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sources":["../../../../src/activities/generateImage/adapter.ts"],"sourcesContent":["import type {\n  ImageGenerationOptions,\n  ImageGenerationResult,\n  ModelInputModalitiesByName,\n} from '../../types'\n\n/**\n * Resolve the size type for a model from the model-size map.\n * If the map has an index signature (i.e. no explicit keys), falls back to string.\n * If the model is an explicit key, uses its mapped size type.\n * Otherwise falls back to string.\n */\n\n/**\n * Configuration for image adapter instances\n */\nexport interface ImageAdapterConfig {\n  apiKey?: string\n  baseUrl?: string\n  timeout?: number\n  maxRetries?: number\n  headers?: Record<string, string>\n}\n\n/**\n * Image adapter interface with pre-resolved generics.\n *\n * An adapter is created by a provider function: `provider('model')` → `adapter`\n * All type resolution happens at the provider call site, not in this interface.\n *\n * Generic parameters:\n * - TModel: The specific model name (e.g., 'dall-e-3')\n * - TProviderOptions: Base provider-specific options (already resolved)\n * - TModelProviderOptionsByName: Map from model name to its specific provider options\n * - TModelSizeByName: Map from model name to its supported sizes\n * - TModelInputModalitiesByName: Map from model name to the non-text prompt\n *   modalities it accepts (constrains the `prompt` part types at compile time)\n */\nexport interface ImageAdapter<\n  TModel extends string = string,\n  TProviderOptions extends object = Record<string, unknown>,\n  TModelProviderOptionsByName extends Record<string, any> = Record<string, any>,\n  TModelSizeByName extends Record<string, string | undefined> = Record<\n    string,\n    string\n  >,\n  TModelInputModalitiesByName extends ModelInputModalitiesByName =\n    ModelInputModalitiesByName,\n> {\n  /** Discriminator for adapter kind - used by generate() to determine API shape */\n  readonly kind: 'image'\n  /** Adapter name identifier */\n  readonly name: string\n  /** The model this adapter is configured for */\n  readonly model: TModel\n\n  /**\n   * @internal Type-only properties for inference. Not assigned at runtime.\n   */\n  '~types': {\n    providerOptions: TProviderOptions\n    modelProviderOptionsByName: TModelProviderOptionsByName\n    modelSizeByName: TModelSizeByName\n    modelInputModalitiesByName: TModelInputModalitiesByName\n  }\n\n  /**\n   * Generate images from a prompt\n   */\n  generateImages: (\n    options: ImageGenerationOptions<TProviderOptions, TModelSizeByName[TModel]>,\n  ) => Promise<ImageGenerationResult>\n}\n\n/**\n * An ImageAdapter with any/unknown type parameters.\n * Useful as a constraint in generic functions and interfaces.\n */\nexport type AnyImageAdapter = ImageAdapter<any, any, any, any, any>\n\n/**\n * Abstract base class for image generation adapters.\n * Extend this class to implement an image adapter for a specific provider.\n *\n * Generic parameters match ImageAdapter - all pre-resolved by the provider function.\n */\nexport abstract class BaseImageAdapter<\n  TModel extends string = string,\n  TProviderOptions extends object = Record<string, unknown>,\n  TModelProviderOptionsByName extends Record<string, any> = Record<string, any>,\n  TModelSizeByName extends Record<string, string | undefined> = Record<\n    string,\n    string\n  >,\n  TModelInputModalitiesByName extends ModelInputModalitiesByName =\n    ModelInputModalitiesByName,\n> implements ImageAdapter<\n  TModel,\n  TProviderOptions,\n  TModelProviderOptionsByName,\n  TModelSizeByName,\n  TModelInputModalitiesByName\n> {\n  readonly kind = 'image' as const\n  abstract readonly name: string\n  readonly model: TModel\n\n  // Type-only property - never assigned at runtime\n  declare '~types': {\n    providerOptions: TProviderOptions\n    modelProviderOptionsByName: TModelProviderOptionsByName\n    modelSizeByName: TModelSizeByName\n    modelInputModalitiesByName: TModelInputModalitiesByName\n  }\n\n  protected config: ImageAdapterConfig\n\n  constructor(model: TModel, config: ImageAdapterConfig = {}) {\n    this.config = config\n    this.model = model\n  }\n\n  abstract generateImages(\n    options: ImageGenerationOptions<TProviderOptions, TModelSizeByName[TModel]>,\n  ): Promise<ImageGenerationResult>\n\n  protected generateId(): string {\n    return `${this.name}-${Date.now()}-${Math.random().toString(36).substring(7)}`\n  }\n}\n"],"names":[],"mappings":"AAsFO,MAAe,iBAgBpB;AAAA,EACS,OAAO;AAAA,EAEP;AAAA,EAUC;AAAA,EAEV,YAAY,OAAe,SAA6B,IAAI;AAC1D,SAAK,SAAS;AACd,SAAK,QAAQ;AAAA,EACf;AAAA,EAMU,aAAqB;AAC7B,WAAO,GAAG,KAAK,IAAI,IAAI,KAAK,KAAK,IAAI,KAAK,OAAA,EAAS,SAAS,EAAE,EAAE,UAAU,CAAC,CAAC;AAAA,EAC9E;AACF;"}

package/dist/esm/activities/generateImage/index.d.ts

@@ -0,0 +1,134 @@
+import { DebugOption } from '../../logger/types.js';
+import { GenerationMiddleware } from '../middleware.js';
+import { ImageAdapter } from './adapter.js';
+import { ImageGenerationResult, MediaPrompt, MediaPromptFor, StreamChunk } from '../../types.js';
+/** The adapter kind this activity handles */
+export declare const kind: "image";
+/**
+ * Extract model-specific provider options from an ImageAdapter via ~types.
+ * If the model has specific options defined in ModelProviderOptions (and not just via index signature),
+ * use those; otherwise fall back to base provider options.
+ */
+export type ImageProviderOptionsForModel<TAdapter, TModel extends string> = TAdapter extends ImageAdapter<any, infer BaseOptions, infer ModelOptions, any> ? string extends keyof ModelOptions ? BaseOptions : TModel extends keyof ModelOptions ? ModelOptions[TModel] : BaseOptions : object;
+/**
+ * Extract model-specific size options from an ImageAdapter via ~types.
+ * If the model has specific sizes defined, use those; otherwise fall back to string.
+ */
+export type ImageSizeForModel<TAdapter, TModel extends string> = TAdapter extends ImageAdapter<any, any, any, infer SizeByName> ? string extends keyof SizeByName ? string : TModel extends keyof SizeByName ? SizeByName[TModel] : string : string;
+/**
+ * Extract the prompt type a model accepts from an ImageAdapter via ~types.
+ * Adapters declare a per-model input-modality map; models in the map get a
+ * `prompt` narrowed to text + their supported part types (text-only models
+ * accept `string | Array<TextPart>`), so unsupported media parts fail at
+ * compile time. Adapters without a map fall back to the full MediaPrompt.
+ */
+export type ImagePromptForModel<TAdapter, TModel extends string> = TAdapter extends ImageAdapter<any, any, any, any, infer ModsByName> ? string extends keyof ModsByName ? MediaPrompt : TModel extends keyof ModsByName ? MediaPromptFor<ModsByName[TModel][number]> : MediaPrompt : MediaPrompt;
+/**
+ * Options for the image activity.
+ * The model is extracted from the adapter's model property.
+ *
+ * @template TAdapter - The image adapter type
+ * @template TStream - Whether to stream the output
+ */
+export type ImageActivityOptions<TAdapter extends ImageAdapter<string, any, any, any>, TStream extends boolean = false> = {
+    /** The image adapter to use (must be created with a model) */
+    adapter: TAdapter & {
+        kind: typeof kind;
+    };
+    /**
+     * Description of the desired image(s). Either a plain string, or — for
+     * models that support image-conditioned generation — an ordered array of
+     * content parts interleaving text with image inputs (image-to-image,
+     * reference-guided, edit, multi-reference). Media parts may carry
+     * `metadata.role` (`'reference' | 'mask' | 'control' | 'character'`) to
+     * disambiguate intent. The accepted part types are narrowed per model via
+     * the adapter's input-modality map.
+     */
+    prompt: ImagePromptForModel<TAdapter, TAdapter['model']>;
+    /** Number of images to generate (default: 1) */
+    numberOfImages?: number;
+    /** Image size in WIDTHxHEIGHT format (e.g., "1024x1024") */
+    size?: ImageSizeForModel<TAdapter, TAdapter['model']>;
+    /**
+     * Whether to stream the image generation result.
+     * When true, returns an AsyncIterable<StreamChunk> for streaming transport.
+     * When false or not provided, returns a Promise<ImageGenerationResult>.
+     *
+     * @default false
+     */
+    stream?: TStream;
+    /**
+     * Enable debug logging. Pass `true` to enable all categories, `false` to
+     * silence everything including errors, or a `DebugConfig` object for granular
+     * control and/or a custom `Logger`.
+     */
+    debug?: DebugOption;
+    /**
+     * Observe-only middleware notified on start, usage, success, and error. Pass
+     * `otelMiddleware()` to emit OpenTelemetry spans, or implement the
+     * `GenerationMiddleware` contract for a custom backend.
+     */
+    middleware?: Array<GenerationMiddleware>;
+} & ({} extends ImageProviderOptionsForModel<TAdapter, TAdapter['model']> ? {
+    /** Provider-specific options for image generation */ modelOptions?: ImageProviderOptionsForModel<TAdapter, TAdapter['model']>;
+} : {
+    /** Provider-specific options for image generation */ modelOptions: ImageProviderOptionsForModel<TAdapter, TAdapter['model']>;
+});
+/**
+ * Result type for the image activity.
+ * - If stream is true: AsyncIterable<StreamChunk>
+ * - Otherwise: Promise<ImageGenerationResult>
+ */
+export type ImageActivityResult<TStream extends boolean = false> = TStream extends true ? AsyncIterable<StreamChunk> : Promise<ImageGenerationResult>;
+/**
+ * Image activity - generates images from text prompts.
+ *
+ * Uses AI image generation models to create images based on natural language descriptions.
+ *
+ * @example Generate a single image
+ * ```ts
+ * import { generateImage } from '@tanstack/ai'
+ * import { openaiImage } from '@tanstack/ai-openai'
+ *
+ * const result = await generateImage({
+ *   adapter: openaiImage('dall-e-3'),
+ *   prompt: 'A serene mountain landscape at sunset'
+ * })
+ *
+ * console.log(result.images[0].url)
+ * ```
+ *
+ * @example Generate multiple images
+ * ```ts
+ * const result = await generateImage({
+ *   adapter: openaiImage('dall-e-2'),
+ *   prompt: 'A cute robot mascot',
+ *   numberOfImages: 4,
+ *   size: '512x512'
+ * })
+ *
+ * result.images.forEach((image, i) => {
+ *   console.log(`Image ${i + 1}: ${image.url}`)
+ * })
+ * ```
+ *
+ * @example With provider-specific options
+ * ```ts
+ * const result = await generateImage({
+ *   adapter: openaiImage('dall-e-3'),
+ *   prompt: 'A professional headshot photo',
+ *   size: '1024x1024',
+ *   modelOptions: {
+ *     quality: 'hd',
+ *     style: 'natural'
+ *   }
+ * })
+ * ```
+ */
+export declare function generateImage<TAdapter extends ImageAdapter<string, any, any, any>, TStream extends boolean = false>(options: ImageActivityOptions<TAdapter, TStream>): ImageActivityResult<TStream>;
+/**
+ * Create typed options for the generateImage() function without executing.
+ */
+export declare function createImageOptions<TAdapter extends ImageAdapter<string, any, any, any>, TStream extends boolean = false>(options: ImageActivityOptions<TAdapter, TStream>): ImageActivityOptions<TAdapter, TStream>;
+export type { ImageAdapter, ImageAdapterConfig, AnyImageAdapter, } from './adapter.js';
+export { BaseImageAdapter } from './adapter.js';

package/dist/esm/activities/generateImage/index.js

@@ -0,0 +1,121 @@
+import { aiEventClient } from "@tanstack/ai-event-client";
+import { streamGenerationResult } from "../stream-generation-result.js";
+import { resolveDebugOption } from "../../logger/resolve.js";
+import { resolveMediaPrompt } from "../../utilities/media-prompt.js";
+import { createGenerationContext, runGenerationStart, runGenerationUsage, runGenerationFinish, runGenerationError } from "../middleware/run.js";
+const kind = "image";
+function createId(prefix) {
+  return `${prefix}-${Date.now()}-${Math.random().toString(36).slice(2, 9)}`;
+}
+function generateImage(options) {
+  if (options.stream) {
+    return streamGenerationResult(
+      () => runGenerateImage(options)
+    );
+  }
+  return runGenerateImage(options);
+}
+async function runGenerateImage(options) {
+  const {
+    adapter,
+    stream: _stream,
+    debug: _debug,
+    middleware,
+    ...rest
+  } = options;
+  const model = adapter.model;
+  const requestId = createId("image");
+  const startTime = Date.now();
+  const logger = resolveDebugOption(options.debug);
+  const mwCtx = createGenerationContext({
+    requestId,
+    activity: "image",
+    provider: adapter.name,
+    model,
+    modelOptions: rest.modelOptions,
+    createId
+  });
+  await runGenerationStart(middleware, mwCtx);
+  const resolved = resolveMediaPrompt(rest.prompt);
+  aiEventClient.emit("image:request:started", {
+    requestId,
+    provider: adapter.name,
+    model,
+    prompt: resolved.text,
+    numberOfImages: rest.numberOfImages,
+    size: rest.size,
+    ...resolved.images.length > 0 && {
+      imageInputCount: resolved.images.length
+    },
+    ...resolved.videos.length > 0 && {
+      videoInputCount: resolved.videos.length
+    },
+    ...resolved.audios.length > 0 && {
+      audioInputCount: resolved.audios.length
+    },
+    modelOptions: rest.modelOptions,
+    timestamp: startTime
+  });
+  logger.request(`activity=generateImage provider=${adapter.name}`, {
+    provider: adapter.name,
+    model
+  });
+  try {
+    const result = await adapter.generateImages({ ...rest, model, logger });
+    const duration = Date.now() - startTime;
+    aiEventClient.emit("image:request:completed", {
+      requestId,
+      provider: adapter.name,
+      model,
+      // GeneratedImage is a discriminated `{ url } | { b64Json }` union, but the
+      // wire shape on the devtools event is a plain optional pair. Use
+      // conditional spreads so the emitted record only sets the field actually
+      // present — `exactOptionalPropertyTypes` rejects `field: undefined`
+      // against `field?: string` targets.
+      images: result.images.map((image) => ({
+        url: image.url,
+        b64Json: image.b64Json
+      })),
+      duration,
+      modelOptions: rest.modelOptions,
+      timestamp: Date.now()
+    });
+    if (result.usage) {
+      aiEventClient.emit("image:usage", {
+        requestId,
+        model,
+        usage: result.usage,
+        modelOptions: rest.modelOptions,
+        timestamp: Date.now()
+      });
+    }
+    logger.output(`activity=generateImage count=${result.images.length}`, {
+      count: result.images.length
+    });
+    if (result.usage) await runGenerationUsage(middleware, mwCtx, result.usage);
+    await runGenerationFinish(middleware, mwCtx, {
+      duration,
+      usage: result.usage
+    });
+    return result;
+  } catch (error) {
+    await runGenerationError(middleware, mwCtx, {
+      error,
+      duration: Date.now() - startTime
+    });
+    logger.errors("generateImage activity failed", {
+      error,
+      source: "generateImage"
+    });
+    throw error;
+  }
+}
+function createImageOptions(options) {
+  return options;
+}
+export {
+  createImageOptions,
+  generateImage,
+  kind
+};
+//# sourceMappingURL=index.js.map

package/dist/esm/activities/generateImage/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sources":["../../../../src/activities/generateImage/index.ts"],"sourcesContent":["/**\n * Image Activity\n *\n * Generates images from text prompts.\n * This is a self-contained module with implementation, types, and JSDoc.\n */\n\nimport { aiEventClient } from '@tanstack/ai-event-client'\nimport { streamGenerationResult } from '../stream-generation-result.js'\nimport { resolveDebugOption } from '../../logger/resolve'\nimport {\n  createGenerationContext,\n  runGenerationError,\n  runGenerationFinish,\n  runGenerationStart,\n  runGenerationUsage,\n} from '../middleware'\nimport { resolveMediaPrompt } from '../../utilities/media-prompt'\nimport type { InternalLogger } from '../../logger/internal-logger'\nimport type { DebugOption } from '../../logger/types'\nimport type { GenerationMiddleware } from '../middleware'\nimport type { ImageAdapter } from './adapter'\nimport type {\n  ImageGenerationResult,\n  MediaPrompt,\n  MediaPromptFor,\n  StreamChunk,\n} from '../../types'\n\n// ===========================\n// Activity Kind\n// ===========================\n\n/** The adapter kind this activity handles */\nexport const kind = 'image' as const\n\n// ===========================\n// Type Extraction Helpers\n// ===========================\n\n/**\n * Extract model-specific provider options from an ImageAdapter via ~types.\n * If the model has specific options defined in ModelProviderOptions (and not just via index signature),\n * use those; otherwise fall back to base provider options.\n */\nexport type ImageProviderOptionsForModel<TAdapter, TModel extends string> =\n  TAdapter extends ImageAdapter<any, infer BaseOptions, infer ModelOptions, any>\n    ? string extends keyof ModelOptions\n      ? // ModelOptions is Record<string, unknown> or has index signature - use BaseOptions\n        BaseOptions\n      : // ModelOptions has explicit keys - check if TModel is one of them\n        TModel extends keyof ModelOptions\n        ? ModelOptions[TModel]\n        : BaseOptions\n    : object\n\n/**\n * Extract model-specific size options from an ImageAdapter via ~types.\n * If the model has specific sizes defined, use those; otherwise fall back to string.\n */\nexport type ImageSizeForModel<TAdapter, TModel extends string> =\n  TAdapter extends ImageAdapter<any, any, any, infer SizeByName>\n    ? string extends keyof SizeByName\n      ? // SizeByName has index signature - fall back to string\n        string\n      : // SizeByName has explicit keys - check if TModel is one of them\n        TModel extends keyof SizeByName\n        ? SizeByName[TModel]\n        : string\n    : string\n\n/**\n * Extract the prompt type a model accepts from an ImageAdapter via ~types.\n * Adapters declare a per-model input-modality map; models in the map get a\n * `prompt` narrowed to text + their supported part types (text-only models\n * accept `string | Array<TextPart>`), so unsupported media parts fail at\n * compile time. Adapters without a map fall back to the full MediaPrompt.\n */\nexport type ImagePromptForModel<TAdapter, TModel extends string> =\n  TAdapter extends ImageAdapter<any, any, any, any, infer ModsByName>\n    ? string extends keyof ModsByName\n      ? // No explicit map - accept the full union\n        MediaPrompt\n      : TModel extends keyof ModsByName\n        ? MediaPromptFor<ModsByName[TModel][number]>\n        : MediaPrompt\n    : MediaPrompt\n\n// ===========================\n// Activity Options Type\n// ===========================\n\n/**\n * Options for the image activity.\n * The model is extracted from the adapter's model property.\n *\n * @template TAdapter - The image adapter type\n * @template TStream - Whether to stream the output\n */\nexport type ImageActivityOptions<\n  TAdapter extends ImageAdapter<string, any, any, any>,\n  TStream extends boolean = false,\n> = {\n  /** The image adapter to use (must be created with a model) */\n  adapter: TAdapter & { kind: typeof kind }\n  /**\n   * Description of the desired image(s). Either a plain string, or — for\n   * models that support image-conditioned generation — an ordered array of\n   * content parts interleaving text with image inputs (image-to-image,\n   * reference-guided, edit, multi-reference). Media parts may carry\n   * `metadata.role` (`'reference' | 'mask' | 'control' | 'character'`) to\n   * disambiguate intent. The accepted part types are narrowed per model via\n   * the adapter's input-modality map.\n   */\n  prompt: ImagePromptForModel<TAdapter, TAdapter['model']>\n  /** Number of images to generate (default: 1) */\n  numberOfImages?: number\n  /** Image size in WIDTHxHEIGHT format (e.g., \"1024x1024\") */\n  size?: ImageSizeForModel<TAdapter, TAdapter['model']>\n  /**\n   * Whether to stream the image generation result.\n   * When true, returns an AsyncIterable<StreamChunk> for streaming transport.\n   * When false or not provided, returns a Promise<ImageGenerationResult>.\n   *\n   * @default false\n   */\n  stream?: TStream\n  /**\n   * Enable debug logging. Pass `true` to enable all categories, `false` to\n   * silence everything including errors, or a `DebugConfig` object for granular\n   * control and/or a custom `Logger`.\n   */\n  debug?: DebugOption\n  /**\n   * Observe-only middleware notified on start, usage, success, and error. Pass\n   * `otelMiddleware()` to emit OpenTelemetry spans, or implement the\n   * `GenerationMiddleware` contract for a custom backend.\n   */\n  middleware?: Array<GenerationMiddleware>\n} & ({} extends ImageProviderOptionsForModel<TAdapter, TAdapter['model']>\n  ? {\n      /** Provider-specific options for image generation */ modelOptions?: ImageProviderOptionsForModel<\n        TAdapter,\n        TAdapter['model']\n      >\n    }\n  : {\n      /** Provider-specific options for image generation */ modelOptions: ImageProviderOptionsForModel<\n        TAdapter,\n        TAdapter['model']\n      >\n    })\n\n// ===========================\n// Activity Result Type\n// ===========================\n\n/**\n * Result type for the image activity.\n * - If stream is true: AsyncIterable<StreamChunk>\n * - Otherwise: Promise<ImageGenerationResult>\n */\nexport type ImageActivityResult<TStream extends boolean = false> =\n  TStream extends true\n    ? AsyncIterable<StreamChunk>\n    : Promise<ImageGenerationResult>\n\nfunction createId(prefix: string): string {\n  return `${prefix}-${Date.now()}-${Math.random().toString(36).slice(2, 9)}`\n}\n\n// ===========================\n// Activity Implementation\n// ===========================\n\n/**\n * Image activity - generates images from text prompts.\n *\n * Uses AI image generation models to create images based on natural language descriptions.\n *\n * @example Generate a single image\n * ```ts\n * import { generateImage } from '@tanstack/ai'\n * import { openaiImage } from '@tanstack/ai-openai'\n *\n * const result = await generateImage({\n *   adapter: openaiImage('dall-e-3'),\n *   prompt: 'A serene mountain landscape at sunset'\n * })\n *\n * console.log(result.images[0].url)\n * ```\n *\n * @example Generate multiple images\n * ```ts\n * const result = await generateImage({\n *   adapter: openaiImage('dall-e-2'),\n *   prompt: 'A cute robot mascot',\n *   numberOfImages: 4,\n *   size: '512x512'\n * })\n *\n * result.images.forEach((image, i) => {\n *   console.log(`Image ${i + 1}: ${image.url}`)\n * })\n * ```\n *\n * @example With provider-specific options\n * ```ts\n * const result = await generateImage({\n *   adapter: openaiImage('dall-e-3'),\n *   prompt: 'A professional headshot photo',\n *   size: '1024x1024',\n *   modelOptions: {\n *     quality: 'hd',\n *     style: 'natural'\n *   }\n * })\n * ```\n */\nexport function generateImage<\n  TAdapter extends ImageAdapter<string, any, any, any>,\n  TStream extends boolean = false,\n>(\n  options: ImageActivityOptions<TAdapter, TStream>,\n): ImageActivityResult<TStream> {\n  if (options.stream) {\n    return streamGenerationResult(() =>\n      runGenerateImage(options),\n    ) as ImageActivityResult<TStream>\n  }\n\n  return runGenerateImage(options) as ImageActivityResult<TStream>\n}\n\n/**\n * Internal implementation of image generation (always non-streaming).\n * Contains all devtools event emission logic.\n */\nasync function runGenerateImage<\n  TAdapter extends ImageAdapter<string, any, any, any>,\n>(\n  options: ImageActivityOptions<TAdapter, boolean>,\n): Promise<ImageGenerationResult> {\n  const {\n    adapter,\n    stream: _stream,\n    debug: _debug,\n    middleware,\n    ...rest\n  } = options\n  const model = adapter.model\n  const requestId = createId('image')\n  const startTime = Date.now()\n  const logger: InternalLogger = resolveDebugOption(options.debug)\n\n  const mwCtx = createGenerationContext({\n    requestId,\n    activity: 'image',\n    provider: adapter.name,\n    model,\n    modelOptions: rest.modelOptions,\n    createId,\n  })\n\n  await runGenerationStart(middleware, mwCtx)\n\n  // Devtools events carry the flattened prompt text plus media-part counts —\n  // the wire payload stays `prompt: string` regardless of the prompt shape.\n  const resolved = resolveMediaPrompt(rest.prompt)\n\n  aiEventClient.emit('image:request:started', {\n    requestId,\n    provider: adapter.name,\n    model,\n    prompt: resolved.text,\n    numberOfImages: rest.numberOfImages,\n    size: rest.size,\n    ...(resolved.images.length > 0 && {\n      imageInputCount: resolved.images.length,\n    }),\n    ...(resolved.videos.length > 0 && {\n      videoInputCount: resolved.videos.length,\n    }),\n    ...(resolved.audios.length > 0 && {\n      audioInputCount: resolved.audios.length,\n    }),\n    modelOptions: rest.modelOptions,\n    timestamp: startTime,\n  })\n\n  logger.request(`activity=generateImage provider=${adapter.name}`, {\n    provider: adapter.name,\n    model,\n  })\n\n  try {\n    const result = await adapter.generateImages({ ...rest, model, logger })\n    const duration = Date.now() - startTime\n\n    aiEventClient.emit('image:request:completed', {\n      requestId,\n      provider: adapter.name,\n      model,\n      // GeneratedImage is a discriminated `{ url } | { b64Json }` union, but the\n      // wire shape on the devtools event is a plain optional pair. Use\n      // conditional spreads so the emitted record only sets the field actually\n      // present — `exactOptionalPropertyTypes` rejects `field: undefined`\n      // against `field?: string` targets.\n      images: result.images.map((image) => ({\n        url: image.url,\n        b64Json: image.b64Json,\n      })),\n      duration,\n      modelOptions: rest.modelOptions,\n      timestamp: Date.now(),\n    })\n\n    if (result.usage) {\n      aiEventClient.emit('image:usage', {\n        requestId,\n        model,\n        usage: result.usage,\n        modelOptions: rest.modelOptions,\n        timestamp: Date.now(),\n      })\n    }\n\n    logger.output(`activity=generateImage count=${result.images.length}`, {\n      count: result.images.length,\n    })\n\n    if (result.usage) await runGenerationUsage(middleware, mwCtx, result.usage)\n    await runGenerationFinish(middleware, mwCtx, {\n      duration,\n      usage: result.usage,\n    })\n\n    return result\n  } catch (error) {\n    await runGenerationError(middleware, mwCtx, {\n      error,\n      duration: Date.now() - startTime,\n    })\n    logger.errors('generateImage activity failed', {\n      error,\n      source: 'generateImage',\n    })\n    throw error\n  }\n}\n\n// ===========================\n// Options Factory\n// ===========================\n\n/**\n * Create typed options for the generateImage() function without executing.\n */\nexport function createImageOptions<\n  TAdapter extends ImageAdapter<string, any, any, any>,\n  TStream extends boolean = false,\n>(\n  options: ImageActivityOptions<TAdapter, TStream>,\n): ImageActivityOptions<TAdapter, TStream> {\n  return options\n}\n\n// Re-export adapter types\nexport type {\n  ImageAdapter,\n  ImageAdapterConfig,\n  AnyImageAdapter,\n} from './adapter'\nexport { BaseImageAdapter } from './adapter'\n"],"names":[],"mappings":";;;;;AAkCO,MAAM,OAAO;AAqIpB,SAAS,SAAS,QAAwB;AACxC,SAAO,GAAG,MAAM,IAAI,KAAK,IAAA,CAAK,IAAI,KAAK,OAAA,EAAS,SAAS,EAAE,EAAE,MAAM,GAAG,CAAC,CAAC;AAC1E;AAmDO,SAAS,cAId,SAC8B;AAC9B,MAAI,QAAQ,QAAQ;AAClB,WAAO;AAAA,MAAuB,MAC5B,iBAAiB,OAAO;AAAA,IAAA;AAAA,EAE5B;AAEA,SAAO,iBAAiB,OAAO;AACjC;AAMA,eAAe,iBAGb,SACgC;AAChC,QAAM;AAAA,IACJ;AAAA,IACA,QAAQ;AAAA,IACR,OAAO;AAAA,IACP;AAAA,IACA,GAAG;AAAA,EAAA,IACD;AACJ,QAAM,QAAQ,QAAQ;AACtB,QAAM,YAAY,SAAS,OAAO;AAClC,QAAM,YAAY,KAAK,IAAA;AACvB,QAAM,SAAyB,mBAAmB,QAAQ,KAAK;AAE/D,QAAM,QAAQ,wBAAwB;AAAA,IACpC;AAAA,IACA,UAAU;AAAA,IACV,UAAU,QAAQ;AAAA,IAClB;AAAA,IACA,cAAc,KAAK;AAAA,IACnB;AAAA,EAAA,CACD;AAED,QAAM,mBAAmB,YAAY,KAAK;AAI1C,QAAM,WAAW,mBAAmB,KAAK,MAAM;AAE/C,gBAAc,KAAK,yBAAyB;AAAA,IAC1C;AAAA,IACA,UAAU,QAAQ;AAAA,IAClB;AAAA,IACA,QAAQ,SAAS;AAAA,IACjB,gBAAgB,KAAK;AAAA,IACrB,MAAM,KAAK;AAAA,IACX,GAAI,SAAS,OAAO,SAAS,KAAK;AAAA,MAChC,iBAAiB,SAAS,OAAO;AAAA,IAAA;AAAA,IAEnC,GAAI,SAAS,OAAO,SAAS,KAAK;AAAA,MAChC,iBAAiB,SAAS,OAAO;AAAA,IAAA;AAAA,IAEnC,GAAI,SAAS,OAAO,SAAS,KAAK;AAAA,MAChC,iBAAiB,SAAS,OAAO;AAAA,IAAA;AAAA,IAEnC,cAAc,KAAK;AAAA,IACnB,WAAW;AAAA,EAAA,CACZ;AAED,SAAO,QAAQ,mCAAmC,QAAQ,IAAI,IAAI;AAAA,IAChE,UAAU,QAAQ;AAAA,IAClB;AAAA,EAAA,CACD;AAED,MAAI;AACF,UAAM,SAAS,MAAM,QAAQ,eAAe,EAAE,GAAG,MAAM,OAAO,QAAQ;AACtE,UAAM,WAAW,KAAK,IAAA,IAAQ;AAE9B,kBAAc,KAAK,2BAA2B;AAAA,MAC5C;AAAA,MACA,UAAU,QAAQ;AAAA,MAClB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,MAMA,QAAQ,OAAO,OAAO,IAAI,CAAC,WAAW;AAAA,QACpC,KAAK,MAAM;AAAA,QACX,SAAS,MAAM;AAAA,MAAA,EACf;AAAA,MACF;AAAA,MACA,cAAc,KAAK;AAAA,MACnB,WAAW,KAAK,IAAA;AAAA,IAAI,CACrB;AAED,QAAI,OAAO,OAAO;AAChB,oBAAc,KAAK,eAAe;AAAA,QAChC;AAAA,QACA;AAAA,QACA,OAAO,OAAO;AAAA,QACd,cAAc,KAAK;AAAA,QACnB,WAAW,KAAK,IAAA;AAAA,MAAI,CACrB;AAAA,IACH;AAEA,WAAO,OAAO,gCAAgC,OAAO,OAAO,MAAM,IAAI;AAAA,MACpE,OAAO,OAAO,OAAO;AAAA,IAAA,CACtB;AAED,QAAI,OAAO,MAAO,OAAM,mBAAmB,YAAY,OAAO,OAAO,KAAK;AAC1E,UAAM,oBAAoB,YAAY,OAAO;AAAA,MAC3C;AAAA,MACA,OAAO,OAAO;AAAA,IAAA,CACf;AAED,WAAO;AAAA,EACT,SAAS,OAAO;AACd,UAAM,mBAAmB,YAAY,OAAO;AAAA,MAC1C;AAAA,MACA,UAAU,KAAK,QAAQ;AAAA,IAAA,CACxB;AACD,WAAO,OAAO,iCAAiC;AAAA,MAC7C;AAAA,MACA,QAAQ;AAAA,IAAA,CACT;AACD,UAAM;AAAA,EACR;AACF;AASO,SAAS,mBAId,SACyC;AACzC,SAAO;AACT;"}

package/dist/esm/activities/generateSpeech/adapter.d.ts

@@ -0,0 +1,62 @@
+import { TTSOptions, TTSResult } from '../../types.js';
+/**
+ * Configuration for TTS adapter instances
+ */
+export interface TTSAdapterConfig {
+    apiKey?: string;
+    baseUrl?: string;
+    timeout?: number;
+    maxRetries?: number;
+    headers?: Record<string, string>;
+}
+/**
+ * TTS adapter interface with pre-resolved generics.
+ *
+ * An adapter is created by a provider function: `provider('model')` → `adapter`
+ * All type resolution happens at the provider call site, not in this interface.
+ *
+ * Generic parameters:
+ * - TModel: The specific model name (e.g., 'tts-1')
+ * - TProviderOptions: Provider-specific options (already resolved)
+ */
+export interface TTSAdapter<TModel extends string = string, TProviderOptions extends object = Record<string, unknown>> {
+    /** Discriminator for adapter kind - used to determine API shape */
+    readonly kind: 'tts';
+    /** Adapter name identifier */
+    readonly name: string;
+    /** The model this adapter is configured for */
+    readonly model: TModel;
+    /**
+     * @internal Type-only properties for inference. Not assigned at runtime.
+     */
+    '~types': {
+        providerOptions: TProviderOptions;
+    };
+    /**
+     * Generate speech from text
+     */
+    generateSpeech: (options: TTSOptions<TProviderOptions>) => Promise<TTSResult>;
+}
+/**
+ * A TTSAdapter with any/unknown type parameters.
+ * Useful as a constraint in generic functions and interfaces.
+ */
+export type AnyTTSAdapter = TTSAdapter<any, any>;
+/**
+ * Abstract base class for text-to-speech adapters.
+ * Extend this class to implement a TTS adapter for a specific provider.
+ *
+ * Generic parameters match TTSAdapter - all pre-resolved by the provider function.
+ */
+export declare abstract class BaseTTSAdapter<TModel extends string = string, TProviderOptions extends object = Record<string, unknown>> implements TTSAdapter<TModel, TProviderOptions> {
+    readonly kind: "tts";
+    abstract readonly name: string;
+    readonly model: TModel;
+    '~types': {
+        providerOptions: TProviderOptions;
+    };
+    protected config: TTSAdapterConfig;
+    constructor(model: TModel, config?: TTSAdapterConfig);
+    abstract generateSpeech(options: TTSOptions<TProviderOptions>): Promise<TTSResult>;
+    protected generateId(): string;
+}

package/dist/esm/activities/generateSpeech/adapter.js

@@ -0,0 +1,16 @@
+class BaseTTSAdapter {
+  kind = "tts";
+  model;
+  config;
+  constructor(model, config = {}) {
+    this.config = config;
+    this.model = model;
+  }
+  generateId() {
+    return `${this.name}-${Date.now()}-${Math.random().toString(36).substring(7)}`;
+  }
+}
+export {
+  BaseTTSAdapter
+};
+//# sourceMappingURL=adapter.js.map

package/dist/esm/activities/generateSpeech/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sources":["../../../../src/activities/generateSpeech/adapter.ts"],"sourcesContent":["import type { TTSOptions, TTSResult } from '../../types'\n\n/**\n * Configuration for TTS adapter instances\n */\nexport interface TTSAdapterConfig {\n  apiKey?: string\n  baseUrl?: string\n  timeout?: number\n  maxRetries?: number\n  headers?: Record<string, string>\n}\n\n/**\n * TTS adapter interface with pre-resolved generics.\n *\n * An adapter is created by a provider function: `provider('model')` → `adapter`\n * All type resolution happens at the provider call site, not in this interface.\n *\n * Generic parameters:\n * - TModel: The specific model name (e.g., 'tts-1')\n * - TProviderOptions: Provider-specific options (already resolved)\n */\nexport interface TTSAdapter<\n  TModel extends string = string,\n  TProviderOptions extends object = Record<string, unknown>,\n> {\n  /** Discriminator for adapter kind - used to determine API shape */\n  readonly kind: 'tts'\n  /** Adapter name identifier */\n  readonly name: string\n  /** The model this adapter is configured for */\n  readonly model: TModel\n\n  /**\n   * @internal Type-only properties for inference. Not assigned at runtime.\n   */\n  '~types': {\n    providerOptions: TProviderOptions\n  }\n\n  /**\n   * Generate speech from text\n   */\n  generateSpeech: (options: TTSOptions<TProviderOptions>) => Promise<TTSResult>\n}\n\n/**\n * A TTSAdapter with any/unknown type parameters.\n * Useful as a constraint in generic functions and interfaces.\n */\nexport type AnyTTSAdapter = TTSAdapter<any, any>\n\n/**\n * Abstract base class for text-to-speech adapters.\n * Extend this class to implement a TTS adapter for a specific provider.\n *\n * Generic parameters match TTSAdapter - all pre-resolved by the provider function.\n */\nexport abstract class BaseTTSAdapter<\n  TModel extends string = string,\n  TProviderOptions extends object = Record<string, unknown>,\n> implements TTSAdapter<TModel, TProviderOptions> {\n  readonly kind = 'tts' as const\n  abstract readonly name: string\n  readonly model: TModel\n\n  // Type-only property - never assigned at runtime\n  declare '~types': {\n    providerOptions: TProviderOptions\n  }\n\n  protected config: TTSAdapterConfig\n\n  constructor(model: TModel, config: TTSAdapterConfig = {}) {\n    this.config = config\n    this.model = model\n  }\n\n  abstract generateSpeech(\n    options: TTSOptions<TProviderOptions>,\n  ): Promise<TTSResult>\n\n  protected generateId(): string {\n    return `${this.name}-${Date.now()}-${Math.random().toString(36).substring(7)}`\n  }\n}\n"],"names":[],"mappings":"AA2DO,MAAe,eAG4B;AAAA,EACvC,OAAO;AAAA,EAEP;AAAA,EAOC;AAAA,EAEV,YAAY,OAAe,SAA2B,IAAI;AACxD,SAAK,SAAS;AACd,SAAK,QAAQ;AAAA,EACf;AAAA,EAMU,aAAqB;AAC7B,WAAO,GAAG,KAAK,IAAI,IAAI,KAAK,KAAK,IAAI,KAAK,OAAA,EAAS,SAAS,EAAE,EAAE,UAAU,CAAC,CAAC;AAAA,EAC9E;AACF;"}

package/dist/esm/activities/generateSpeech/index.d.ts

@@ -0,0 +1,96 @@
+import { DebugOption } from '../../logger/types.js';
+import { GenerationMiddleware } from '../middleware.js';
+import { TTSAdapter } from './adapter.js';
+import { StreamChunk, TTSResult } from '../../types.js';
+/** The adapter kind this activity handles */
+export declare const kind: "tts";
+/**
+ * Extract provider options from a TTSAdapter via ~types.
+ */
+export type TTSProviderOptions<TAdapter> = TAdapter extends TTSAdapter<any, any> ? TAdapter['~types']['providerOptions'] : object;
+/**
+ * Options for the TTS activity.
+ * The model is extracted from the adapter's model property.
+ *
+ * @template TAdapter - The TTS adapter type
+ * @template TStream - Whether to stream the output
+ */
+export interface TTSActivityOptions<TAdapter extends TTSAdapter<string, TTSProviderOptions<TAdapter>>, TStream extends boolean = false> {
+    /** The TTS adapter to use (must be created with a model) */
+    adapter: TAdapter & {
+        kind: typeof kind;
+    };
+    /** The text to convert to speech */
+    text: string;
+    /** The voice to use for generation */
+    voice?: string;
+    /** The output audio format */
+    format?: 'mp3' | 'opus' | 'aac' | 'flac' | 'wav' | 'pcm';
+    /** The speed of the generated audio (0.25 to 4.0) */
+    speed?: number;
+    /** Provider-specific options for TTS generation */
+    modelOptions?: TTSProviderOptions<TAdapter>;
+    /**
+     * Whether to stream the generation result.
+     * When true, returns an AsyncIterable<StreamChunk> for streaming transport.
+     * When false or not provided, returns a Promise<TTSResult>.
+     *
+     * @default false
+     */
+    stream?: TStream;
+    /**
+     * Enable debug logging. Pass `true` to enable all categories, `false` to
+     * silence everything including errors, or a `DebugConfig` object for granular
+     * control and/or a custom `Logger`.
+     */
+    debug?: DebugOption;
+    /**
+     * Observe-only middleware notified on start, usage, success, and error. Pass
+     * `otelMiddleware()` to emit OpenTelemetry spans, or implement the
+     * `GenerationMiddleware` contract for a custom backend.
+     */
+    middleware?: Array<GenerationMiddleware>;
+}
+/**
+ * Result type for the TTS activity.
+ * - If stream is true: AsyncIterable<StreamChunk>
+ * - Otherwise: Promise<TTSResult>
+ */
+export type TTSActivityResult<TStream extends boolean = false> = TStream extends true ? AsyncIterable<StreamChunk> : Promise<TTSResult>;
+/**
+ * TTS activity - generates speech from text.
+ *
+ * Uses AI text-to-speech models to create audio from natural language text.
+ *
+ * @example Generate speech from text
+ * ```ts
+ * import { generateSpeech } from '@tanstack/ai'
+ * import { openaiSpeech } from '@tanstack/ai-openai'
+ *
+ * const result = await generateSpeech({
+ *   adapter: openaiSpeech('tts-1-hd'),
+ *   text: 'Hello, welcome to TanStack AI!',
+ *   voice: 'nova'
+ * })
+ *
+ * console.log(result.audio) // base64-encoded audio
+ * ```
+ *
+ * @example With format and speed options
+ * ```ts
+ * const result = await generateSpeech({
+ *   adapter: openaiSpeech('tts-1'),
+ *   text: 'This is slower speech.',
+ *   voice: 'alloy',
+ *   format: 'wav',
+ *   speed: 0.8
+ * })
+ * ```
+ */
+export declare function generateSpeech<TAdapter extends TTSAdapter<string, TTSProviderOptions<TAdapter>>, TStream extends boolean = false>(options: TTSActivityOptions<TAdapter, TStream>): TTSActivityResult<TStream>;
+/**
+ * Create typed options for the generateSpeech() function without executing.
+ */
+export declare function createSpeechOptions<TAdapter extends TTSAdapter<string, TTSProviderOptions<TAdapter>>, TStream extends boolean = false>(options: TTSActivityOptions<TAdapter, TStream>): TTSActivityOptions<TAdapter, TStream>;
+export type { TTSAdapter, TTSAdapterConfig, AnyTTSAdapter } from './adapter.js';
+export { BaseTTSAdapter } from './adapter.js';