@h-ai/ai 0.1.0-alpha.41 → 0.1.0-alpha.43

package/README.md

@@ -1,6 +1,6 @@
 # @h-ai/ai
 
-AI 能力模块，提供统一的 `ai` 服务对象，覆盖 LLM 对话、工具调用、MCP、Embedding、记忆、检索/RAG、知识库、上下文管理、文件解析、Rerank、语音（ASR/TTS）与 A2A。Node.js 侧通过 `ai.init()` 初始化，浏览器侧通过 API/client 代理访问。
+AI 能力模块，提供统一的 `ai` 服务对象，覆盖 LLM 对话、文生图、工具调用、MCP、Embedding、记忆、检索/RAG、知识库、上下文管理、文件解析、Rerank、语音（ASR/TTS）与 A2A。Node.js 侧通过 `ai.init()` 初始化，浏览器侧通过 API/client 代理访问。
 
 ## 能力概览
 
@@ -16,6 +16,7 @@ AI 能力模块，提供统一的 `ai` 服务对象，覆盖 LLM 对话、工具
 - `ai.context`：LLM + Memory + RAG + 压缩的一体化会话管理。
 - `ai.file` / `ai.rerank`：文件解析/OCR 与文本重排序。
 - `ai.audio`：语音识别（ASR）与语音合成（TTS），支持完整与流式调用，覆盖 OpenAI / MiMo / Qwen / 豆包平台。
+- `ai.image`：文生图，覆盖 OpenAI GPT Image、Google Gemini Image / Nano Banana、Qwen-Image 2.0/3.0、Seedream 4.x/5.x 与 Pollinations 免费额度模型。
 - `ai.a2a`：Agent-to-Agent 请求处理与远端调用。
 - `@h-ai/ai/client`：前端轻量客户端（配合 API 服务）。
 - `AIStoreProvider`：统一存储抽象；无数据库时默认使用进程内临时 Store，已初始化 reldb + vecdb 时自动使用持久化 DB Provider。
@@ -210,11 +211,15 @@ if (setup.success) {
 
 ### 语音（Audio）
 
-先在 `ai.init()` 中注册语音模型并映射默认识别/合成模型（凭据可回退到平台环境变量）：
+先在 `ai.init()` 中注册语音模型并映射默认识别/合成模型。凭据默认回退到平台环境变量；只有 LLM 与语音模型确认使用同一凭据时，才显式启用 `inheritLlmApiKey`：
 
 ```ts
 await ai.init({
+  llm: {
+    apiKey: process.env.HAI_AI_LLM_API_KEY,
+  },
   audio: {
+    inheritLlmApiKey: true,
     models: [
       { id: 'asr', provider: 'qwen', model: 'qwen3-asr-flash-realtime', operations: ['transcribe'] },
       { id: 'tts', provider: 'qwen', model: 'qwen3-tts-flash-realtime', operations: ['synthesize'] },
@@ -261,12 +266,44 @@ const caps = ai.audio.getCapabilities({ operation: 'synthesize', model: 'tts' })
 if (caps.success && caps.data.synthesize?.streamingAudioOutput) { /* 可实时 TTS */ }
 ```
 
+`OptionalSecretSchema` 统一用于 LLM、Audio 与 Image 的可选密钥字段：YAML `null`、空字符串和纯空白字符串都会规范化为 `undefined`。语音密钥优先级为模型条目 `apiKey` → 启用继承后的 LLM `apiKey` → 对应平台环境变量；`inheritLlmApiKey` 默认关闭，避免跨供应商误用密钥。
+
 > `synthesizeStream` 严格按 `segment_started → audio* → segment_done` 产出事件。`segment_started` 携带服务端解析 Provider 后的**真实输出音频参数**（`format` / `sampleRate` / `channels`），播放器据此正确解码，不应按请求参数猜测格式。播放器只有在对应音频真正播放完成后才应把该段文本计入 `spokenText`；播放状态仍由应用管理。
 
 取消/超时/连接错误统一为领域错误：`AbortSignal` 触发 → `AUDIO_CANCELLED`（超时 → `AUDIO_TIMEOUT`），连接失败或 `end` 前异常断连 → `AUDIO_CONNECTION_FAILED`。实时连接时长受 `audio.maxStreamDurationMs`（默认 5 分钟）限制。
 
 浏览器 / 移动端通过 `@h-ai/serv` 暴露的统一语音 WebSocket 入口访问，`@h-ai/ai/client` 提供与 Node 端一致的 `audio.*` API（传输细节内部隐藏）。浏览器客户端严格区分正常结束、取消（`AUDIO_CANCELLED`）与异常断连（`AUDIO_CONNECTION_FAILED`）：取消或在 `end` 前断连会抛出对应领域错误码，`synthesize` 不会把未完成的部分音频当作成功结果返回。
 
+### 文生图（Image）
+
+在 `ai.init()` 注册模型后，调用方只接收标准化图片字节，不感知厂商 Base64、内联数据或临时 URL：
+
+```ts
+await ai.init({
+  image: {
+    models: [
+      { id: 'image', provider: 'openai', model: 'gpt-image-2' },
+      { id: 'free', provider: 'pollinations', model: 'zimage' },
+    ],
+    generateModel: 'image',
+  },
+})
+
+const result = await ai.image.generate({
+  prompt: '一个正在构建开源框架的友好机器人',
+  size: { width: 1024, height: 1024 },
+  referenceImages: [
+    { data: await readImage('character.png'), mimeType: 'image/png' },
+  ],
+})
+if (result.success) {
+  const { data, mimeType } = result.data.images[0]!
+  await saveImage(data, mimeType)
+}
+```
+
+`referenceImages` 可省略；公共层只接受图片字节与 MIME，multipart、`inlineData`、Data URL 等差异由 Provider 转换。凭据可在模型条目传入，或使用 `HAI_AI_IMAGE_<PROVIDER>_API_KEY` / 厂商环境变量。Qwen 与 Seedream 返回的临时 URL 会在 Provider 内立即下载，不会泄漏到公共返回值。完整接口差异与模型说明见 [REFERENCE.md](./REFERENCE.md#image)。
+
 ### Context 管理器
 
 ```ts
@@ -460,6 +497,7 @@ if (!result.success) {
 - `hai:ai:600-701`：Retrieval/RAG。
 - `hai:ai:800-805`：Knowledge。
 - `hai:ai:050-059`：Audio。
+- `hai:ai:060-065`：Image。
 - `hai:ai:900-905`：Memory。
 - `hai:ai:980-984`：A2A。
 

package/dist/{ai-audio-ws-protocol-DE4WxxHM.d.ts → ai-audio-ws-protocol-CUIXMEkt.d.ts}

@@ -1,6 +1,6 @@
 import * as _h_ai_core from '@h-ai/core';
 import { HaiResult } from '@h-ai/core';
-import { aG as ChatMessage, af as InteractionScope, O as MemoryType, br as RagOptions, bv as ReasoningOptions, c2 as ToolRegistryOperations, bd as MemoryEntry, bK as SessionInfo, b6 as LLMOperations, bj as MemoryOperations, bq as RagOperations, bu as ReasoningOperations, c as AIConfig, d as AIConfigInput, av as AIStoreProvider, c4 as ToolsOperations, bO as StreamOperations, bB as RetrievalOperations, b0 as KnowledgeOperations, ap as A2AOperations, q as AudioOperations, l as AudioFormat } from './ai-reasoning-types-B8PH0w2_.js';
+import { aP as ChatMessage, ao as InteractionScope, U as MemoryType, bA as RagOptions, bE as ReasoningOptions, cb as ToolRegistryOperations, bm as MemoryEntry, bT as SessionInfo, bf as LLMOperations, bs as MemoryOperations, bz as RagOperations, bD as ReasoningOperations, c as AIConfig, d as AIConfigInput, aE as AIStoreProvider, cd as ToolsOperations, bX as StreamOperations, bK as RetrievalOperations, b9 as KnowledgeOperations, ay as A2AOperations, q as AudioOperations, l as AudioFormat } from './ai-reasoning-types-DsVJ4CE8.js';
 import * as zod from 'zod';
 import { z } from 'zod';
 import { Buffer } from 'node:buffer';
@@ -918,6 +918,63 @@ interface FileOperations {
     parseText: (content: Buffer | string, filename?: string) => Promise<HaiResult<string>>;
 }
 
+/**
+ * @h-ai/ai — 文生图公共类型
+ *
+ * 公共契约只暴露业务需要的提示词、尺寸和标准化图片字节。
+ * 厂商请求字段、临时 URL 与响应中间态由 Provider 内部处理。
+ * @module image/ai-image-types
+ */
+
+/** 图片像素尺寸 */
+interface ImageSize {
+    /** 宽度（像素） */
+    width: number;
+    /** 高度（像素） */
+    height: number;
+}
+/** 可选参考图；Provider 会转换为对应厂商的传输格式 */
+interface ReferenceImage {
+    /** 图片二进制内容 */
+    data: Uint8Array;
+    /** 图片 MIME 类型，例如 `image/png`、`image/jpeg` */
+    mimeType: string;
+}
+/** 文生图请求 */
+interface GenerateImageRequest {
+    /** 图片内容与风格提示词 */
+    prompt: string;
+    /** 模型 ID；不传时使用配置中的默认文生图模型 */
+    model?: string;
+    /** 输出像素尺寸；不传时由模型决定 */
+    size?: ImageSize;
+    /** 用于图生图、编辑或风格参考的一张或多张图片 */
+    referenceImages?: ReferenceImage[];
+    /** 请求取消信号 */
+    signal?: AbortSignal;
+}
+/** 标准化后的单张图片 */
+interface GeneratedImage {
+    /** 图片二进制内容 */
+    data: Uint8Array;
+    /** 图片 MIME 类型 */
+    mimeType: string;
+    /** 实际宽度（厂商返回时提供） */
+    width?: number;
+    /** 实际高度（厂商返回时提供） */
+    height?: number;
+}
+/** 文生图结果 */
+interface GenerateImageResult {
+    /** 厂商实际返回的图片；通常为一张 */
+    images: GeneratedImage[];
+}
+/** 文生图操作接口（通过 `ai.image` 访问） */
+interface ImageOperations {
+    /** 根据文本提示生成图片 */
+    generate: (request: GenerateImageRequest) => Promise<HaiResult<GenerateImageResult>>;
+}
+
 /**
  * @h-ai/ai — MCP 子功能类型
  *
@@ -1414,6 +1471,12 @@ declare const HaiAIError: {
     readonly AUDIO_TIMEOUT: _h_ai_core.HaiErrorDef;
     readonly AUDIO_INPUT_TOO_LARGE: _h_ai_core.HaiErrorDef;
     readonly AUDIO_CANCELLED: _h_ai_core.HaiErrorDef;
+    readonly IMAGE_INVALID_REQUEST: _h_ai_core.HaiErrorDef;
+    readonly IMAGE_MODEL_NOT_FOUND: _h_ai_core.HaiErrorDef;
+    readonly IMAGE_PROVIDER_NOT_FOUND: _h_ai_core.HaiErrorDef;
+    readonly IMAGE_UPSTREAM_ERROR: _h_ai_core.HaiErrorDef;
+    readonly IMAGE_PROTOCOL_ERROR: _h_ai_core.HaiErrorDef;
+    readonly IMAGE_CANCELLED: _h_ai_core.HaiErrorDef;
     readonly CONTEXT_COMPRESS_FAILED: _h_ai_core.HaiErrorDef;
     readonly CONTEXT_SUMMARIZE_FAILED: _h_ai_core.HaiErrorDef;
     readonly CONTEXT_TOKEN_ESTIMATE_FAILED: _h_ai_core.HaiErrorDef;
@@ -1522,6 +1585,8 @@ interface AIFunctions {
     readonly a2a: A2AOperations;
     /** Audio 操作（语音识别 / 语音合成），需要先调用 `init()` 并配置 `audio` */
     readonly audio: AudioOperations;
+    /** Image 操作（文生图），需要先调用 `init()` 并配置 `image` */
+    readonly image: ImageOperations;
 }
 
 /** 统一语音入口的默认路径（相对 API 前缀） */
@@ -1689,4 +1754,4 @@ interface AudioWsEndMessage {
 /** 服务端 JSON 消息（合成音频以二进制帧返回，不走 JSON） */
 type AudioWsServerMessage = AudioWsSpeechMessage | AudioWsTranscriptMessage | AudioWsSegmentStartedMessage | AudioWsSegmentDoneMessage | AudioWsErrorMessage | AudioWsEndMessage;
 
-export { type MCPPromptArgument as $, type AIFunctions as A, type ContextChatOptions as B, type CompressionStrategy as C, type ContextChatResult as D, type ContextDeps as E, type ContextManager as F, type ContextManagerOptions as G, HaiAIError as H, type ContextOperations as I, type ContextResetOptions as J, type ContextStreamEvent as K, type ConversationTurn as L, type McpServerOptions as M, type ConversationTurnStatus as N, type EmbeddingItem as O, type EmbeddingOperations as P, type EmbeddingProvider as Q, type EmbeddingRequest as R, type EmbeddingResponse as S, type FileOperations as T, type FileParseMethod as U, type FileParseOptions as V, type FileParseRequest as W, type FileParseResult as X, type MCPContext as Y, type MCPOperations as Z, type MCPPrompt as _, type AIInitOptions as a, type MCPPromptContent as a0, type MCPPromptMessage as a1, type MCPProvider as a2, type MCPResource as a3, type MCPResourceContent as a4, type MCPToolDefinition as a5, type MCPToolHandler as a6, type OutputFormat as a7, type PersonaOperations as a8, type PersonaProfile as a9, type PersonaProfileInput as aa, type PersonaProfileUpdate as ab, type PersonaScopeOptions as ac, type RerankDocument as ad, type RerankItem as ae, type RerankOperations as af, type RerankRequest as ag, type RerankResponse as ah, type SummaryOperations as ai, type SummaryOptions as aj, type SummaryResult as ak, type TokenOperations as al, AUDIO_WS_PATH as b, AudioFormatSchema as c, type AudioWsClientMessage as d, AudioWsClientMessageSchema as e, type AudioWsDoneMessage as f, AudioWsDoneMessageSchema as g, type AudioWsEndMessage as h, type AudioWsErrorMessage as i, type AudioWsSegmentDoneMessage as j, type AudioWsSegmentStartedMessage as k, type AudioWsServerMessage as l, type AudioWsSpeechMessage as m, type AudioWsStartMessage as n, AudioWsStartMessageSchema as o, type AudioWsTextMessage as p, AudioWsTextMessageSchema as q, type AudioWsTranscriptMessage as r, CompressionStrategySchema as s, type AIMCPFunctionsDeps as t, type CommitTurnInput as u, type CompressOperations as v, type CompressOptions as w, type CompressResult as x, type ConsolidateOptions as y, type ConsolidateResult as z };
+export { type FileParseOptions as $, type AIFunctions as A, type CompressResult as B, type CompressionStrategy as C, type ConsolidateOptions as D, type ConsolidateResult as E, type ContextChatOptions as F, type GenerateImageRequest as G, HaiAIError as H, type ImageOperations as I, type ContextChatResult as J, type ContextDeps as K, type ContextManager as L, type McpServerOptions as M, type ContextManagerOptions as N, type ContextOperations as O, type ContextResetOptions as P, type ContextStreamEvent as Q, type ReferenceImage as R, type ConversationTurn as S, type ConversationTurnStatus as T, type EmbeddingItem as U, type EmbeddingOperations as V, type EmbeddingProvider as W, type EmbeddingRequest as X, type EmbeddingResponse as Y, type FileOperations as Z, type FileParseMethod as _, type AIInitOptions as a, type FileParseRequest as a0, type FileParseResult as a1, type MCPContext as a2, type MCPOperations as a3, type MCPPrompt as a4, type MCPPromptArgument as a5, type MCPPromptContent as a6, type MCPPromptMessage as a7, type MCPProvider as a8, type MCPResource as a9, type MCPResourceContent as aa, type MCPToolDefinition as ab, type MCPToolHandler as ac, type OutputFormat as ad, type PersonaOperations as ae, type PersonaProfile as af, type PersonaProfileInput as ag, type PersonaProfileUpdate as ah, type PersonaScopeOptions as ai, type RerankDocument as aj, type RerankItem as ak, type RerankOperations as al, type RerankRequest as am, type RerankResponse as an, type SummaryOperations as ao, type SummaryOptions as ap, type SummaryResult as aq, type TokenOperations as ar, AUDIO_WS_PATH as b, AudioFormatSchema as c, type AudioWsClientMessage as d, AudioWsClientMessageSchema as e, type AudioWsDoneMessage as f, AudioWsDoneMessageSchema as g, type AudioWsEndMessage as h, type AudioWsErrorMessage as i, type AudioWsSegmentDoneMessage as j, type AudioWsSegmentStartedMessage as k, type AudioWsServerMessage as l, type AudioWsSpeechMessage as m, type AudioWsStartMessage as n, AudioWsStartMessageSchema as o, type AudioWsTextMessage as p, AudioWsTextMessageSchema as q, type AudioWsTranscriptMessage as r, CompressionStrategySchema as s, type GenerateImageResult as t, type GeneratedImage as u, type ImageSize as v, type AIMCPFunctionsDeps as w, type CommitTurnInput as x, type CompressOperations as y, type CompressOptions as z };

package/dist/{ai-reasoning-types-B8PH0w2_.d.ts → ai-reasoning-types-DsVJ4CE8.d.ts}

@@ -2214,6 +2214,13 @@ interface MemoryOperations extends MemoryCoreOperations {
  * @module ai-config
  */
 
+/**
+ * 可选密钥 Schema。
+ *
+ * 配置文件中的空插值、空白字符串或 YAML null 都表示“未配置”，输出统一为 undefined；
+ * 非空密钥保持原值，避免模块调用方重复编写兼容清洗逻辑。
+ */
+declare const OptionalSecretSchema: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
 /**
  * 模型场景枚举
  *
@@ -2285,7 +2292,7 @@ declare const ModelEntrySchema: z.ZodObject<{
         responses: "responses";
         anthropic: "anthropic";
     }>>;
-    apiKey: z.ZodOptional<z.ZodString>;
+    apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
     baseUrl: z.ZodOptional<z.ZodURL>;
     maxTokens: z.ZodOptional<z.ZodNumber>;
     temperature: z.ZodOptional<z.ZodNumber>;
@@ -2325,7 +2332,7 @@ type ModelEntry = z.infer<typeof ModelEntrySchema>;
  * ```
  */
 declare const LLMConfigSchema: z.ZodObject<{
-    apiKey: z.ZodOptional<z.ZodString>;
+    apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
     baseUrl: z.ZodOptional<z.ZodURL>;
     model: z.ZodDefault<z.ZodOptional<z.ZodString>>;
     api: z.ZodDefault<z.ZodOptional<z.ZodEnum<{
@@ -2345,7 +2352,7 @@ declare const LLMConfigSchema: z.ZodObject<{
             responses: "responses";
             anthropic: "anthropic";
         }>>;
-        apiKey: z.ZodOptional<z.ZodString>;
+        apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
         baseUrl: z.ZodOptional<z.ZodURL>;
         maxTokens: z.ZodOptional<z.ZodNumber>;
         temperature: z.ZodOptional<z.ZodNumber>;
@@ -2787,10 +2794,10 @@ declare const AudioModelEntrySchema: z.ZodObject<{
     }>;
     model: z.ZodString;
     operations: z.ZodUnion<readonly [z.ZodTuple<[z.ZodLiteral<"transcribe">], null>, z.ZodTuple<[z.ZodLiteral<"synthesize">], null>, z.ZodTuple<[z.ZodLiteral<"transcribe">, z.ZodLiteral<"synthesize">], null>]>;
-    apiKey: z.ZodOptional<z.ZodString>;
+    apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
     baseUrl: z.ZodOptional<z.ZodString>;
-    appKey: z.ZodOptional<z.ZodString>;
-    accessKey: z.ZodOptional<z.ZodString>;
+    appKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
+    accessKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
     resourceId: z.ZodOptional<z.ZodString>;
     workspaceId: z.ZodOptional<z.ZodString>;
     timeout: z.ZodOptional<z.ZodNumber>;
@@ -2827,14 +2834,15 @@ declare const AudioConfigSchema: z.ZodObject<{
         }>;
         model: z.ZodString;
         operations: z.ZodUnion<readonly [z.ZodTuple<[z.ZodLiteral<"transcribe">], null>, z.ZodTuple<[z.ZodLiteral<"synthesize">], null>, z.ZodTuple<[z.ZodLiteral<"transcribe">, z.ZodLiteral<"synthesize">], null>]>;
-        apiKey: z.ZodOptional<z.ZodString>;
+        apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
         baseUrl: z.ZodOptional<z.ZodString>;
-        appKey: z.ZodOptional<z.ZodString>;
-        accessKey: z.ZodOptional<z.ZodString>;
+        appKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
+        accessKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
         resourceId: z.ZodOptional<z.ZodString>;
         workspaceId: z.ZodOptional<z.ZodString>;
         timeout: z.ZodOptional<z.ZodNumber>;
     }, z.core.$strip>>>;
+    inheritLlmApiKey: z.ZodDefault<z.ZodBoolean>;
     transcribeModel: z.ZodOptional<z.ZodString>;
     synthesizeModel: z.ZodOptional<z.ZodString>;
     maxAudioBytes: z.ZodDefault<z.ZodNumber>;
@@ -2845,7 +2853,8 @@ type AudioConfig = z.infer<typeof AudioConfigSchema>;
 /**
  * 已解析的语音模型配置
  *
- * 由 `resolveAudioModel()` 返回，凭据已合并环境变量、端点已应用平台默认值。
+ * 由 `resolveAudioModel()` 返回，凭据已按模型条目、显式 LLM 继承、平台环境变量的顺序解析，
+ * 端点已应用平台默认值。
  */
 interface ResolvedAudioModel {
     /** 模型条目 ID */
@@ -2854,7 +2863,7 @@ interface ResolvedAudioModel {
     provider: AudioProviderName;
     /** 厂商模型名 */
     model: string;
-    /** API Key（条目 > 平台环境变量） */
+    /** API Key（模型条目 > 显式启用的 LLM 密钥继承 > 平台环境变量） */
     apiKey: string | undefined;
     /** 端点（条目 > 平台默认） */
     baseUrl: string;
@@ -2877,9 +2886,85 @@ interface ResolvedAudioModel {
  * @param audioConfig - Audio 配置
  * @param operation - 操作类型（识别 / 合成），决定使用哪个默认模型
  * @param explicit - 请求显式指定的模型 ID（最高优先级）
+ * @param llmApiKey - LLM 全局密钥，仅在 `inheritLlmApiKey` 启用时参与解析
  * @returns 成功返回已解析模型；无匹配模型返回 `AUDIO_MODEL_NOT_FOUND`；缺少凭据返回 `CONFIGURATION_ERROR`
  */
-declare function resolveAudioModel(audioConfig: AudioConfig, operation: 'transcribe' | 'synthesize', explicit?: string): HaiResult<ResolvedAudioModel>;
+declare function resolveAudioModel(audioConfig: AudioConfig, operation: 'transcribe' | 'synthesize', explicit?: string, llmApiKey?: string): HaiResult<ResolvedAudioModel>;
+/**
+ * 文生图平台枚举
+ *
+ * - `openai` — OpenAI GPT Image（Image API）
+ * - `google` — Google Gemini Image / Nano Banana（Generate Content API）
+ * - `qwen` — 阿里云百炼 Qwen-Image 2.0 / 3.0
+ * - `seedream` — 火山方舟 Seedream 4.x / 5.x
+ * - `pollinations` — Pollinations 免费额度图片 API
+ */
+declare const ImageProviderSchema: z.ZodEnum<{
+    openai: "openai";
+    qwen: "qwen";
+    google: "google";
+    seedream: "seedream";
+    pollinations: "pollinations";
+}>;
+/** 文生图平台类型 */
+type ImageProviderName = z.infer<typeof ImageProviderSchema>;
+/** 文生图模型条目 Schema */
+declare const ImageModelEntrySchema: z.ZodObject<{
+    id: z.ZodString;
+    provider: z.ZodEnum<{
+        openai: "openai";
+        qwen: "qwen";
+        google: "google";
+        seedream: "seedream";
+        pollinations: "pollinations";
+    }>;
+    model: z.ZodString;
+    apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
+    baseUrl: z.ZodOptional<z.ZodURL>;
+    workspaceId: z.ZodOptional<z.ZodString>;
+    timeout: z.ZodDefault<z.ZodNumber>;
+}, z.core.$strip>;
+/** 文生图模型条目类型 */
+type ImageModelEntry = z.infer<typeof ImageModelEntrySchema>;
+/** 文生图配置 Schema */
+declare const ImageConfigSchema: z.ZodObject<{
+    models: z.ZodOptional<z.ZodArray<z.ZodObject<{
+        id: z.ZodString;
+        provider: z.ZodEnum<{
+            openai: "openai";
+            qwen: "qwen";
+            google: "google";
+            seedream: "seedream";
+            pollinations: "pollinations";
+        }>;
+        model: z.ZodString;
+        apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
+        baseUrl: z.ZodOptional<z.ZodURL>;
+        workspaceId: z.ZodOptional<z.ZodString>;
+        timeout: z.ZodDefault<z.ZodNumber>;
+    }, z.core.$strip>>>;
+    generateModel: z.ZodOptional<z.ZodString>;
+}, z.core.$strip>;
+/** 文生图配置类型 */
+type ImageConfig = z.infer<typeof ImageConfigSchema>;
+/** 已解析的文生图模型 */
+interface ResolvedImageModel {
+    id: string;
+    provider: ImageProviderName;
+    model: string;
+    apiKey: string;
+    baseUrl: string;
+    workspaceId?: string;
+    timeout: number;
+}
+/**
+ * 解析文生图模型配置
+ *
+ * @param imageConfig - 文生图配置
+ * @param explicit - 请求显式模型 ID 或厂商模型名
+ * @returns 已解析模型；模型或凭据缺失时返回 HaiResult 错误
+ */
+declare function resolveImageModel(imageConfig: ImageConfig, explicit?: string): HaiResult<ResolvedImageModel>;
 /**
  * AI 配置 Schema
  *
@@ -2925,7 +3010,7 @@ declare function resolveAudioModel(audioConfig: AudioConfig, operation: 'transcr
  */
 declare const AIConfigSchema: z.ZodObject<{
     llm: z.ZodDefault<z.ZodObject<{
-        apiKey: z.ZodOptional<z.ZodString>;
+        apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
         baseUrl: z.ZodOptional<z.ZodURL>;
         model: z.ZodDefault<z.ZodOptional<z.ZodString>>;
         api: z.ZodDefault<z.ZodOptional<z.ZodEnum<{
@@ -2945,7 +3030,7 @@ declare const AIConfigSchema: z.ZodObject<{
                 responses: "responses";
                 anthropic: "anthropic";
             }>>;
-            apiKey: z.ZodOptional<z.ZodString>;
+            apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
             baseUrl: z.ZodOptional<z.ZodURL>;
             maxTokens: z.ZodOptional<z.ZodNumber>;
             temperature: z.ZodOptional<z.ZodNumber>;
@@ -3093,19 +3178,38 @@ declare const AIConfigSchema: z.ZodObject<{
             }>;
             model: z.ZodString;
             operations: z.ZodUnion<readonly [z.ZodTuple<[z.ZodLiteral<"transcribe">], null>, z.ZodTuple<[z.ZodLiteral<"synthesize">], null>, z.ZodTuple<[z.ZodLiteral<"transcribe">, z.ZodLiteral<"synthesize">], null>]>;
-            apiKey: z.ZodOptional<z.ZodString>;
+            apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
             baseUrl: z.ZodOptional<z.ZodString>;
-            appKey: z.ZodOptional<z.ZodString>;
-            accessKey: z.ZodOptional<z.ZodString>;
+            appKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
+            accessKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
             resourceId: z.ZodOptional<z.ZodString>;
             workspaceId: z.ZodOptional<z.ZodString>;
             timeout: z.ZodOptional<z.ZodNumber>;
         }, z.core.$strip>>>;
+        inheritLlmApiKey: z.ZodDefault<z.ZodBoolean>;
         transcribeModel: z.ZodOptional<z.ZodString>;
         synthesizeModel: z.ZodOptional<z.ZodString>;
         maxAudioBytes: z.ZodDefault<z.ZodNumber>;
         maxStreamDurationMs: z.ZodDefault<z.ZodNumber>;
     }, z.core.$strip>>;
+    image: z.ZodOptional<z.ZodObject<{
+        models: z.ZodOptional<z.ZodArray<z.ZodObject<{
+            id: z.ZodString;
+            provider: z.ZodEnum<{
+                openai: "openai";
+                qwen: "qwen";
+                google: "google";
+                seedream: "seedream";
+                pollinations: "pollinations";
+            }>;
+            model: z.ZodString;
+            apiKey: z.ZodOptional<z.ZodPipe<z.ZodNullable<z.ZodString>, z.ZodTransform<string | undefined, string | null>>>;
+            baseUrl: z.ZodOptional<z.ZodURL>;
+            workspaceId: z.ZodOptional<z.ZodString>;
+            timeout: z.ZodDefault<z.ZodNumber>;
+        }, z.core.$strip>>>;
+        generateModel: z.ZodOptional<z.ZodString>;
+    }, z.core.$strip>>;
 }, z.core.$strip>;
 /** AI 配置类型（校验后的完整类型） */
 type AIConfig = z.infer<typeof AIConfigSchema>;
@@ -3643,4 +3747,4 @@ interface ReasoningOperations {
     runStream: (query: string, options?: ReasoningOptions) => AsyncIterable<ReasoningStreamEvent>;
 }
 
-export { type SummaryConfig as $, type A2AConfig as A, MCPConfigSchema as B, type CompressConfig as C, type MCPServerCapabilities as D, type EmbeddingConfig as E, type FileConfig as F, MCPServerCapabilitiesSchema as G, type MCPServerConfig as H, MCPServerConfigSchema as I, type MemoryConfig as J, type KnowledgeConfig as K, type LLMConfig as L, type MCPConfig as M, MemoryConfigSchema as N, type MemoryType as O, MemoryTypeSchema as P, type ModelEntry as Q, ModelEntrySchema as R, type ModelScenario as S, ModelScenarioSchema as T, type ResolveRequiredModelEntryOptions as U, type ResolvedAudioModel as V, type ResolvedModelConfig as W, type RetrievalConfig as X, RetrievalConfigSchema as Y, type RetrievalSourceConfig as Z, RetrievalSourceSchema as _, A2AConfigSchema as a, type KnowledgeIngestResult as a$, SummaryConfigSchema as a0, type SynthesisEvent as a1, type SynthesisRequest as a2, type SynthesisResult as a3, type SynthesisStreamRequest as a4, type SynthesisTextSegment as a5, type TokenConfig as a6, TokenConfigSchema as a7, type TranscriptionEvent as a8, type TranscriptionRequest as a9, type ChatCompletionChoice as aA, type ChatCompletionChunk as aB, type ChatCompletionDelta as aC, type ChatCompletionRequest as aD, type ChatCompletionResponse as aE, type ChatHistoryOptions as aF, type ChatMessage as aG, type ChatRecord as aH, type Citation as aI, type DefineToolOptions as aJ, type DeveloperMessage as aK, type EntityDocumentRelation as aL, type EntityDocumentResult as aM, type EntityListOptions as aN, type EntityQueryOptions as aO, type GenerateObjectRequest as aP, type ImageContent as aQ, type KnowledgeAskOptions as aR, type KnowledgeAskResult as aS, type KnowledgeDocumentInfo as aT, type KnowledgeDocumentListOptions as aU, type KnowledgeDocumentRemoveOptions as aV, type KnowledgeEntity as aW, type KnowledgeIngestBatchProgress as aX, type KnowledgeIngestBatchResult as aY, type KnowledgeIngestFileInput as aZ, type KnowledgeIngestInput as a_, type TranscriptionResult as aa, type TranscriptionStreamRequest as ab, resolveAudioModel as ac, resolveModelApi as ad, resolveModelEntry as ae, type InteractionScope as af, type A2AAgentCardConfig as ag, type A2AApiKeySecurity as ah, type A2AAuthenticator as ai, type A2ACallOptions as aj, type A2ACallResult as ak, type A2ACallerIdentity as al, type A2AContextInfo as am, type A2AHandleResult as an, type A2AMessageRecord as ao, type A2AOperations as ap, type A2ASecurityConfig as aq, type A2ATaskFilter as ar, type AILLMFunctionsDeps as as, type AIRelStore as at, type AIRelStoreOptions as au, type AIStoreProvider as av, type AIVectorBackend as aw, type AIVectorStore as ax, type AskOptions as ay, type AssistantMessage as az, A2ASkillConfigSchema as b, type ToolExecutionContext as b$, type KnowledgeOperations as b0, type KnowledgeRetrieveItem as b1, type KnowledgeRetrieveOptions as b2, type KnowledgeRetrieveResult as b3, type KnowledgeSetupOptions as b4, type KnowledgeStore as b5, type LLMOperations as b6, type LLMProvider as b7, type MemoryAccessScope as b8, type MemoryAdminOperations as b9, type ReasoningStreamEvent as bA, type RetrievalOperations as bB, type RetrievalRequest as bC, type RetrievalResult as bD, type RetrievalResultItem as bE, type RetrievalSource as bF, type SSEDecoder as bG, type SSEEvent as bH, type ScopedMemoryBinding as bI, type ScopedMemoryOperations as bJ, type SessionInfo as bK, type StoreFilter as bL, type StorePage as bM, type StoreScope as bN, type StreamOperations as bO, type StreamProcessor as bP, type StreamResult as bQ, type SystemMessage as bR, type TempModelConfig as bS, type TextContent as bT, type TokenUsage as bU, type Tool as bV, type ToolAuthorizationRequest as bW, type ToolAuthorizer as bX, type ToolCall as bY, type ToolDefinition as bZ, type ToolErrorType as b_, type MemoryClearAllOptions as ba, type MemoryClearOptions as bb, type MemoryCoreOperations as bc, type MemoryEntry as bd, type MemoryEntryInput as be, type MemoryExtractOptions as bf, type MemoryInjectionOptions as bg, type MemoryListOptions as bh, type MemoryListPageOptions as bi, type MemoryOperations as bj, type MemoryRecallOptions as bk, type MemoryUpdateInput as bl, type MessageContent as bm, type MessageRole as bn, type ObjectRef as bo, type RagContextItem as bp, type RagOperations as bq, type RagOptions as br, type RagResult as bs, type RagStreamEvent as bt, type ReasoningOperations as bu, type ReasoningOptions as bv, type ReasoningResult as bw, type ReasoningStep as bx, type ReasoningStepType as by, type ReasoningStrategy as bz, type AIConfig as c, type ToolExecutionOptions as c0, type ToolMessage as c1, type ToolRegistryOperations as c2, type ToolRegistryOptions as c3, type ToolsOperations as c4, type UserMessage as c5, type WhereClause as c6, type WhereOperator as c7, type WhereValue as c8, type AIConfigInput as d, AIConfigSchema as e, type ApiType as f, ApiTypeSchema as g, type AudioCapabilitiesRequest as h, type AudioConfig as i, AudioConfigSchema as j, type AudioContent as k, type AudioFormat as l, type AudioInputStream as m, type AudioModelCapabilities as n, type AudioModelEntry as o, AudioModelEntrySchema as p, type AudioOperations as q, type AudioProviderName as r, AudioProviderSchema as s, CompressConfigSchema as t, EmbeddingConfigSchema as u, type EntityType as v, EntityTypeSchema as w, FileConfigSchema as x, KnowledgeConfigSchema as y, LLMConfigSchema as z };
+export { type ResolveRequiredModelEntryOptions as $, type A2AConfig as A, ImageModelEntrySchema as B, type CompressConfig as C, type ImageProviderName as D, type EmbeddingConfig as E, type FileConfig as F, ImageProviderSchema as G, KnowledgeConfigSchema as H, type ImageConfig as I, LLMConfigSchema as J, type KnowledgeConfig as K, type LLMConfig as L, type MCPConfig as M, MCPConfigSchema as N, type MCPServerCapabilities as O, MCPServerCapabilitiesSchema as P, type MCPServerConfig as Q, MCPServerConfigSchema as R, type MemoryConfig as S, MemoryConfigSchema as T, type MemoryType as U, MemoryTypeSchema as V, type ModelEntry as W, ModelEntrySchema as X, type ModelScenario as Y, ModelScenarioSchema as Z, OptionalSecretSchema as _, A2AConfigSchema as a, type KnowledgeAskResult as a$, type ResolvedAudioModel as a0, type ResolvedImageModel as a1, type ResolvedModelConfig as a2, type RetrievalConfig as a3, RetrievalConfigSchema as a4, type RetrievalSourceConfig as a5, RetrievalSourceSchema as a6, type SummaryConfig as a7, SummaryConfigSchema as a8, type SynthesisEvent as a9, type A2ATaskFilter as aA, type AILLMFunctionsDeps as aB, type AIRelStore as aC, type AIRelStoreOptions as aD, type AIStoreProvider as aE, type AIVectorBackend as aF, type AIVectorStore as aG, type AskOptions as aH, type AssistantMessage as aI, type ChatCompletionChoice as aJ, type ChatCompletionChunk as aK, type ChatCompletionDelta as aL, type ChatCompletionRequest as aM, type ChatCompletionResponse as aN, type ChatHistoryOptions as aO, type ChatMessage as aP, type ChatRecord as aQ, type Citation as aR, type DefineToolOptions as aS, type DeveloperMessage as aT, type EntityDocumentRelation as aU, type EntityDocumentResult as aV, type EntityListOptions as aW, type EntityQueryOptions as aX, type GenerateObjectRequest as aY, type ImageContent as aZ, type KnowledgeAskOptions as a_, type SynthesisRequest as aa, type SynthesisResult as ab, type SynthesisStreamRequest as ac, type SynthesisTextSegment as ad, type TokenConfig as ae, TokenConfigSchema as af, type TranscriptionEvent as ag, type TranscriptionRequest as ah, type TranscriptionResult as ai, type TranscriptionStreamRequest as aj, resolveAudioModel as ak, resolveImageModel as al, resolveModelApi as am, resolveModelEntry as an, type InteractionScope as ao, type A2AAgentCardConfig as ap, type A2AApiKeySecurity as aq, type A2AAuthenticator as ar, type A2ACallOptions as as, type A2ACallResult as at, type A2ACallerIdentity as au, type A2AContextInfo as av, type A2AHandleResult as aw, type A2AMessageRecord as ax, type A2AOperations as ay, type A2ASecurityConfig as az, A2ASkillConfigSchema as b, type TempModelConfig as b$, type KnowledgeDocumentInfo as b0, type KnowledgeDocumentListOptions as b1, type KnowledgeDocumentRemoveOptions as b2, type KnowledgeEntity as b3, type KnowledgeIngestBatchProgress as b4, type KnowledgeIngestBatchResult as b5, type KnowledgeIngestFileInput as b6, type KnowledgeIngestInput as b7, type KnowledgeIngestResult as b8, type KnowledgeOperations as b9, type RagOptions as bA, type RagResult as bB, type RagStreamEvent as bC, type ReasoningOperations as bD, type ReasoningOptions as bE, type ReasoningResult as bF, type ReasoningStep as bG, type ReasoningStepType as bH, type ReasoningStrategy as bI, type ReasoningStreamEvent as bJ, type RetrievalOperations as bK, type RetrievalRequest as bL, type RetrievalResult as bM, type RetrievalResultItem as bN, type RetrievalSource as bO, type SSEDecoder as bP, type SSEEvent as bQ, type ScopedMemoryBinding as bR, type ScopedMemoryOperations as bS, type SessionInfo as bT, type StoreFilter as bU, type StorePage as bV, type StoreScope as bW, type StreamOperations as bX, type StreamProcessor as bY, type StreamResult as bZ, type SystemMessage as b_, type KnowledgeRetrieveItem as ba, type KnowledgeRetrieveOptions as bb, type KnowledgeRetrieveResult as bc, type KnowledgeSetupOptions as bd, type KnowledgeStore as be, type LLMOperations as bf, type LLMProvider as bg, type MemoryAccessScope as bh, type MemoryAdminOperations as bi, type MemoryClearAllOptions as bj, type MemoryClearOptions as bk, type MemoryCoreOperations as bl, type MemoryEntry as bm, type MemoryEntryInput as bn, type MemoryExtractOptions as bo, type MemoryInjectionOptions as bp, type MemoryListOptions as bq, type MemoryListPageOptions as br, type MemoryOperations as bs, type MemoryRecallOptions as bt, type MemoryUpdateInput as bu, type MessageContent as bv, type MessageRole as bw, type ObjectRef as bx, type RagContextItem as by, type RagOperations as bz, type AIConfig as c, type TextContent as c0, type TokenUsage as c1, type Tool as c2, type ToolAuthorizationRequest as c3, type ToolAuthorizer as c4, type ToolCall as c5, type ToolDefinition as c6, type ToolErrorType as c7, type ToolExecutionContext as c8, type ToolExecutionOptions as c9, type ToolMessage as ca, type ToolRegistryOperations as cb, type ToolRegistryOptions as cc, type ToolsOperations as cd, type UserMessage as ce, type WhereClause as cf, type WhereOperator as cg, type WhereValue as ch, type AIConfigInput as d, AIConfigSchema as e, type ApiType as f, ApiTypeSchema as g, type AudioCapabilitiesRequest as h, type AudioConfig as i, AudioConfigSchema as j, type AudioContent as k, type AudioFormat as l, type AudioInputStream as m, type AudioModelCapabilities as n, type AudioModelEntry as o, AudioModelEntrySchema as p, type AudioOperations as q, type AudioProviderName as r, AudioProviderSchema as s, CompressConfigSchema as t, EmbeddingConfigSchema as u, type EntityType as v, EntityTypeSchema as w, FileConfigSchema as x, ImageConfigSchema as y, type ImageModelEntry as z };

package/dist/browser.d.ts

@@ -1,5 +1,5 @@
-export { A as A2AConfig, a as A2AConfigSchema, b as A2ASkillConfigSchema, c as AIConfig, d as AIConfigInput, e as AIConfigSchema, f as ApiType, g as ApiTypeSchema, h as AudioCapabilitiesRequest, i as AudioConfig, j as AudioConfigSchema, k as AudioContent, l as AudioFormat, m as AudioInputStream, n as AudioModelCapabilities, o as AudioModelEntry, p as AudioModelEntrySchema, q as AudioOperations, r as AudioProviderName, s as AudioProviderSchema, C as CompressConfig, t as CompressConfigSchema, E as EmbeddingConfig, u as EmbeddingConfigSchema, v as EntityType, w as EntityTypeSchema, F as FileConfig, x as FileConfigSchema, K as KnowledgeConfig, y as KnowledgeConfigSchema, L as LLMConfig, z as LLMConfigSchema, M as MCPConfig, B as MCPConfigSchema, D as MCPServerCapabilities, G as MCPServerCapabilitiesSchema, H as MCPServerConfig, I as MCPServerConfigSchema, J as MemoryConfig, N as MemoryConfigSchema, O as MemoryType, P as MemoryTypeSchema, Q as ModelEntry, R as ModelEntrySchema, S as ModelScenario, T as ModelScenarioSchema, U as ResolveRequiredModelEntryOptions, V as ResolvedAudioModel, W as ResolvedModelConfig, X as RetrievalConfig, Y as RetrievalConfigSchema, Z as RetrievalSourceConfig, _ as RetrievalSourceSchema, $ as SummaryConfig, a0 as SummaryConfigSchema, a1 as SynthesisEvent, a2 as SynthesisRequest, a3 as SynthesisResult, a4 as SynthesisStreamRequest, a5 as SynthesisTextSegment, a6 as TokenConfig, a7 as TokenConfigSchema, a8 as TranscriptionEvent, a9 as TranscriptionRequest, aa as TranscriptionResult, ab as TranscriptionStreamRequest, ac as resolveAudioModel, ad as resolveModelApi, ae as resolveModelEntry } from './ai-reasoning-types-B8PH0w2_.js';
-export { A as AIFunctions, a as AIInitOptions, b as AUDIO_WS_PATH, c as AudioFormatSchema, d as AudioWsClientMessage, e as AudioWsClientMessageSchema, f as AudioWsDoneMessage, g as AudioWsDoneMessageSchema, h as AudioWsEndMessage, i as AudioWsErrorMessage, j as AudioWsSegmentDoneMessage, k as AudioWsSegmentStartedMessage, l as AudioWsServerMessage, m as AudioWsSpeechMessage, n as AudioWsStartMessage, o as AudioWsStartMessageSchema, p as AudioWsTextMessage, q as AudioWsTextMessageSchema, r as AudioWsTranscriptMessage, C as CompressionStrategy, s as CompressionStrategySchema, H as HaiAIError } from './ai-audio-ws-protocol-DE4WxxHM.js';
+export { A as A2AConfig, a as A2AConfigSchema, b as A2ASkillConfigSchema, c as AIConfig, d as AIConfigInput, e as AIConfigSchema, f as ApiType, g as ApiTypeSchema, h as AudioCapabilitiesRequest, i as AudioConfig, j as AudioConfigSchema, k as AudioContent, l as AudioFormat, m as AudioInputStream, n as AudioModelCapabilities, o as AudioModelEntry, p as AudioModelEntrySchema, q as AudioOperations, r as AudioProviderName, s as AudioProviderSchema, C as CompressConfig, t as CompressConfigSchema, E as EmbeddingConfig, u as EmbeddingConfigSchema, v as EntityType, w as EntityTypeSchema, F as FileConfig, x as FileConfigSchema, I as ImageConfig, y as ImageConfigSchema, z as ImageModelEntry, B as ImageModelEntrySchema, D as ImageProviderName, G as ImageProviderSchema, K as KnowledgeConfig, H as KnowledgeConfigSchema, L as LLMConfig, J as LLMConfigSchema, M as MCPConfig, N as MCPConfigSchema, O as MCPServerCapabilities, P as MCPServerCapabilitiesSchema, Q as MCPServerConfig, R as MCPServerConfigSchema, S as MemoryConfig, T as MemoryConfigSchema, U as MemoryType, V as MemoryTypeSchema, W as ModelEntry, X as ModelEntrySchema, Y as ModelScenario, Z as ModelScenarioSchema, _ as OptionalSecretSchema, $ as ResolveRequiredModelEntryOptions, a0 as ResolvedAudioModel, a1 as ResolvedImageModel, a2 as ResolvedModelConfig, a3 as RetrievalConfig, a4 as RetrievalConfigSchema, a5 as RetrievalSourceConfig, a6 as RetrievalSourceSchema, a7 as SummaryConfig, a8 as SummaryConfigSchema, a9 as SynthesisEvent, aa as SynthesisRequest, ab as SynthesisResult, ac as SynthesisStreamRequest, ad as SynthesisTextSegment, ae as TokenConfig, af as TokenConfigSchema, ag as TranscriptionEvent, ah as TranscriptionRequest, ai as TranscriptionResult, aj as TranscriptionStreamRequest, ak as resolveAudioModel, al as resolveImageModel, am as resolveModelApi, an as resolveModelEntry } from './ai-reasoning-types-DsVJ4CE8.js';
+export { A as AIFunctions, a as AIInitOptions, b as AUDIO_WS_PATH, c as AudioFormatSchema, d as AudioWsClientMessage, e as AudioWsClientMessageSchema, f as AudioWsDoneMessage, g as AudioWsDoneMessageSchema, h as AudioWsEndMessage, i as AudioWsErrorMessage, j as AudioWsSegmentDoneMessage, k as AudioWsSegmentStartedMessage, l as AudioWsServerMessage, m as AudioWsSpeechMessage, n as AudioWsStartMessage, o as AudioWsStartMessageSchema, p as AudioWsTextMessage, q as AudioWsTextMessageSchema, r as AudioWsTranscriptMessage, C as CompressionStrategy, s as CompressionStrategySchema, G as GenerateImageRequest, t as GenerateImageResult, u as GeneratedImage, H as HaiAIError, I as ImageOperations, v as ImageSize, R as ReferenceImage } from './ai-audio-ws-protocol-CUIXMEkt.js';
 export { A2AClientOperations, AIApiAdapter, AIClient, AIClientConfig, AudioClientConfig, AudioClientOperations, AudioTicketRequest, StreamOptions, StreamProgress, collectStreamContent, createA2AClient, createAIClient, createAudioClient, createUnconfiguredAudioClient, parseSSE } from './client/index.js';
 import '@a2a-js/sdk/server';
 import '@h-ai/core';

package/dist/browser.js

@@ -1,5 +1,5 @@
-export { A2AConfigSchema, A2ASkillConfigSchema, AIConfigSchema, AUDIO_WS_PATH, ApiTypeSchema, AudioConfigSchema, AudioFormatSchema, AudioModelEntrySchema, AudioProviderSchema, AudioWsClientMessageSchema, AudioWsDoneMessageSchema, AudioWsStartMessageSchema, AudioWsTextMessageSchema, CompressConfigSchema, CompressionStrategySchema, EmbeddingConfigSchema, EntityTypeSchema, FileConfigSchema, KnowledgeConfigSchema, LLMConfigSchema, MCPConfigSchema, MCPServerCapabilitiesSchema, MCPServerConfigSchema, MemoryConfigSchema, MemoryTypeSchema, ModelEntrySchema, ModelScenarioSchema, RetrievalConfigSchema, RetrievalSourceSchema, SummaryConfigSchema, TokenConfigSchema, resolveAudioModel, resolveModelApi, resolveModelEntry } from './chunk-DRSZFSCT.js';
-export { collectStreamContent, createA2AClient, createAIClient, createAudioClient, createUnconfiguredAudioClient, parseSSE } from './chunk-ELH2IPOE.js';
-export { HaiAIError } from './chunk-P3GND76X.js';
+export { A2AConfigSchema, A2ASkillConfigSchema, AIConfigSchema, AUDIO_WS_PATH, ApiTypeSchema, AudioConfigSchema, AudioFormatSchema, AudioModelEntrySchema, AudioProviderSchema, AudioWsClientMessageSchema, AudioWsDoneMessageSchema, AudioWsStartMessageSchema, AudioWsTextMessageSchema, CompressConfigSchema, CompressionStrategySchema, EmbeddingConfigSchema, EntityTypeSchema, FileConfigSchema, ImageConfigSchema, ImageModelEntrySchema, ImageProviderSchema, KnowledgeConfigSchema, LLMConfigSchema, MCPConfigSchema, MCPServerCapabilitiesSchema, MCPServerConfigSchema, MemoryConfigSchema, MemoryTypeSchema, ModelEntrySchema, ModelScenarioSchema, OptionalSecretSchema, RetrievalConfigSchema, RetrievalSourceSchema, SummaryConfigSchema, TokenConfigSchema, resolveAudioModel, resolveImageModel, resolveModelApi, resolveModelEntry } from './chunk-E3W2H6FL.js';
+export { collectStreamContent, createA2AClient, createAIClient, createAudioClient, createUnconfiguredAudioClient, parseSSE } from './chunk-6RWS5OWJ.js';
+export { HaiAIError } from './chunk-JC7QBT3C.js';
 //# sourceMappingURL=browser.js.map
 //# sourceMappingURL=browser.js.map

package/dist/{chunk-ELH2IPOE.js → chunk-6RWS5OWJ.js}

@@ -1,4 +1,4 @@
-import { HaiAIError } from './chunk-P3GND76X.js';
+import { HaiAIError } from './chunk-JC7QBT3C.js';
 
 // src/client/ai-a2a-client.ts
 var A2A_PATH = {
@@ -574,5 +574,5 @@ async function collectStreamContent(stream) {
 }
 
 export { collectStreamContent, createA2AClient, createAIClient, createAudioClient, createUnconfiguredAudioClient, parseSSE };
-//# sourceMappingURL=chunk-ELH2IPOE.js.map
-//# sourceMappingURL=chunk-ELH2IPOE.js.map
+//# sourceMappingURL=chunk-6RWS5OWJ.js.map
+//# sourceMappingURL=chunk-6RWS5OWJ.js.map