modular-voice-agent-sdk 2.9.1 → 2.9.2

package/dist/backends/agent/index.js

@@ -6,9 +6,9 @@
  * only receives text responses and tool activity notifications.
  */
 // Core
-export { AgentLLM } from './llm';
+export { AgentLLM } from './llm.js';
 // Providers
-export { OpenCodeAgentProvider } from './opencode-provider';
-export { PiAgentProvider } from './pi-agent-provider';
-export { PiCodingAgentProvider } from './pi-coding-agent-provider';
+export { OpenCodeAgentProvider } from './opencode-provider.js';
+export { PiAgentProvider } from './pi-agent-provider.js';
+export { PiCodingAgentProvider } from './pi-coding-agent-provider.js';
 //# sourceMappingURL=index.js.map

package/dist/backends/cloud/audio-llm.js

@@ -12,7 +12,7 @@
  *
  * Works with: OpenAI gpt-audio-mini, gpt-audio, and other audio-capable OpenAI-compatible endpoints.
  */
-import { LLMLogger, LLMConversationTracker } from '../../services';
+import { LLMLogger, LLMConversationTracker } from '../../services/index.js';
 export class CloudAudioLLM {
     config;
     ready = false;

package/dist/backends/cloud/index.js

@@ -2,9 +2,9 @@
  * Cloud Backend - Multi-provider API support
  * Works with: OpenAI, Anthropic, Ollama, vLLM, LMStudio, and any compatible endpoint
  */
-export { CloudLLM } from './llm';
+export { CloudLLM } from './llm.js';
 // Audio LLM (multimodal - implements both STTPipeline and LLMPipeline)
-export { CloudAudioLLM } from './audio-llm';
-export { OpenAICompletionsProvider } from './openai-provider';
-export { AnthropicMessagesProvider } from './anthropic-provider';
+export { CloudAudioLLM } from './audio-llm.js';
+export { OpenAICompletionsProvider } from './openai-provider.js';
+export { AnthropicMessagesProvider } from './anthropic-provider.js';
 //# sourceMappingURL=index.js.map

package/dist/backends/cloud/llm.js

@@ -5,8 +5,8 @@
  * Uses native fetch with streaming - no external dependencies required.
  * Supports native tool calling via provider-specific APIs.
  */
-import { LLMLogger, LLMConversationTracker } from '../../services';
-import { OpenAICompletionsProvider } from './openai-provider';
+import { LLMLogger, LLMConversationTracker } from '../../services/index.js';
+import { OpenAICompletionsProvider } from './openai-provider.js';
 export class CloudLLM {
     config;
     provider;

package/dist/backends/index.js

@@ -2,5 +2,5 @@
 // Native backends must be imported directly: import { ... } from 'modular-voice-agent-sdk/native'
 // Cloud backends must be imported directly: import { CloudLLM } from 'modular-voice-agent-sdk/cloud'
 // Web Speech APIs are in the client module: import { WebSpeechSTT, WebSpeechTTS } from 'modular-voice-agent-sdk/client'
-export * from './transformers';
+export * from './transformers/index.js';
 //# sourceMappingURL=index.js.map

package/dist/backends/native/audio-llm.js

@@ -16,7 +16,7 @@ import { spawn } from 'child_process';
 import { existsSync, writeFileSync, unlinkSync } from 'fs';
 import { tmpdir } from 'os';
 import { join } from 'path';
-import { LLMLogger, LLMConversationTracker } from '../../services';
+import { LLMLogger, LLMConversationTracker } from '../../services.js';
 export class NativeAudioLLM {
     config;
     ready = false;

package/dist/backends/native/index.js

@@ -1,7 +1,7 @@
-export { NativeSTT } from './stt';
-export { NativeLLM } from './llm';
-export { NativeTTS, synthesizeToWav } from './tts';
-export { PiperTTSProvider, KokoroTTSProvider } from './tts-providers';
+export { NativeSTT } from './stt.js';
+export { NativeLLM } from './llm.js';
+export { NativeTTS, synthesizeToWav } from './tts.js';
+export { PiperTTSProvider, KokoroTTSProvider } from './tts-providers.js';
 // Cache utilities (Node.js only)
-export { getCacheDir, getModelsDir, getBinDir, getModelPath, getBinaryPath, defaultBinaries, getCacheStatus, checkModelsInstalled, } from '../../cache';
+export { getCacheDir, getModelsDir, getBinDir, getModelPath, getBinaryPath, defaultBinaries, getCacheStatus, checkModelsInstalled, } from '../../cache.js';
 //# sourceMappingURL=index.js.map

package/dist/backends/native/llm.js

@@ -13,7 +13,7 @@ import { spawn } from 'child_process';
 import { existsSync, writeFileSync, unlinkSync } from 'fs';
 import { tmpdir } from 'os';
 import { join } from 'path';
-import { LLMLogger, LLMConversationTracker } from '../../services';
+import { LLMLogger, LLMConversationTracker } from '../../services/index.js';
 export class NativeLLM {
     config;
     ready = false;

package/dist/backends/native/tts.js

@@ -9,10 +9,10 @@ import { execSync } from 'child_process';
 import { existsSync, readFileSync, unlinkSync } from 'fs';
 import { join } from 'path';
 import { tmpdir } from 'os';
-import { BufferedAudioPlayable } from '../../types';
-import { PiperTTSProvider } from './tts-providers';
-import { getBinaryPath } from '../../cache';
-import { encodeWav } from '../../server/encoding';
+import { BufferedAudioPlayable } from '../../types.js';
+import { PiperTTSProvider } from './tts-providers.js';
+import { getBinaryPath } from '../../cache.js';
+import { encodeWav } from '../../server/encoding.js';
 export class NativeTTS {
     config;
     provider;

package/dist/backends/transformers/index.js

@@ -1,4 +1,4 @@
-export { TransformersSTT } from './stt';
-export { TransformersLLM } from './llm';
-export { TransformersTTS } from './tts';
+export { TransformersSTT } from './stt.js';
+export { TransformersLLM } from './llm.js';
+export { TransformersTTS } from './tts.js';
 //# sourceMappingURL=index.js.map

package/dist/backends/transformers/llm.js

@@ -10,7 +10,7 @@
  * for JSON-based tool calling, parsed by VoicePipeline.
  */
 import { pipeline } from '@huggingface/transformers';
-import { LLMLogger, LLMConversationTracker } from '../../services';
+import { LLMLogger, LLMConversationTracker } from '../../services/index.js';
 export class TransformersLLM {
     config;
     modelStore;

package/dist/backends/transformers/tts.js

@@ -5,7 +5,7 @@
  * Supports SpeechT5 and other TTS models from Hugging Face.
  */
 import { pipeline } from '@huggingface/transformers';
-import { BufferedAudioPlayable } from '../../types';
+import { BufferedAudioPlayable } from '../../types.js';
 export class TransformersTTS {
     config;
     modelStore;

package/dist/cli.js

@@ -11,8 +11,8 @@
  */
 import { readdirSync, statSync } from 'fs';
 import { join } from 'path';
-import { getCacheDir, getModelsDir } from './cache';
-import { setup, setupBinaries } from './setup';
+import { getCacheDir, getModelsDir } from './cache.js';
+import { setup, setupBinaries } from './setup.js';
 // ============ CLI Progress Reporter ============
 function cliProgressReporter(event) {
     switch (event.type) {

package/dist/client/index.js

@@ -21,15 +21,15 @@
  * ```
  */
 // Main SDK
-export { VoiceClient, createVoiceClient } from './voice-client';
+export { VoiceClient, createVoiceClient } from './voice-client.js';
 // Browser speech APIs (for local STT/TTS)
-export { WebSpeechSTT } from './web-speech-stt';
-export { WebSpeechTTS } from './web-speech-tts';
-export { WebSocketTransport } from './transports/websocket';
-export { HttpSseTransport } from './transports/http-sse';
+export { WebSpeechSTT } from './web-speech-stt.js';
+export { WebSpeechTTS } from './web-speech-tts.js';
+export { WebSocketTransport } from './transports/websocket.js';
+export { HttpSseTransport } from './transports/http-sse.js';
 // Lower-level utilities (for custom implementations)
-export { AudioRecorder } from './audio-recorder';
-export { AudioPlayer } from './audio-player';
+export { AudioRecorder } from './audio-recorder.js';
+export { AudioPlayer } from './audio-player.js';
 // Protocol types and utilities
-export { float32ToBase64, base64ToFloat32, generateId, } from './protocol';
+export { float32ToBase64, base64ToFloat32, generateId, } from './protocol.js';
 //# sourceMappingURL=index.js.map

package/dist/client/transports/index.js

@@ -1,3 +1,3 @@
-export { WebSocketTransport } from './websocket';
-export { HttpSseTransport } from './http-sse';
+export { WebSocketTransport } from './websocket.js';
+export { HttpSseTransport } from './http-sse.js';
 //# sourceMappingURL=index.js.map

package/dist/client/voice-client.js

@@ -17,13 +17,13 @@
  * - Custom: Pass any Transport implementation via the transport option
  * - Built-in alternatives: HttpSseTransport (HTTP POST + SSE)
  */
-import { VoicePipeline } from '../voice-pipeline';
-import { AudioRecorder } from './audio-recorder';
-import { AudioPlayer } from './audio-player';
-import { WebSpeechSTT } from './web-speech-stt';
-import { WebSpeechTTS } from './web-speech-tts';
-import { float32ToBase64, base64ToFloat32, generateId, } from './protocol';
-import { WebSocketTransport } from './transports/websocket';
+import { VoicePipeline } from '../voice-pipeline.js';
+import { AudioRecorder } from './audio-recorder.js';
+import { AudioPlayer } from './audio-player.js';
+import { WebSpeechSTT } from './web-speech-stt.js';
+import { WebSpeechTTS } from './web-speech-tts.js';
+import { float32ToBase64, base64ToFloat32, generateId, } from './protocol.js';
+import { WebSocketTransport } from './transports/websocket.js';
 // ============ Helpers ============
 function isWebSpeechSTT(obj) {
     return obj instanceof WebSpeechSTT;

package/dist/index.js

@@ -3,11 +3,11 @@
  * Isomorphic STT → LLM → TTS pipeline
  */
 // Main orchestrator
-export { VoicePipeline, createVoicePipeline } from './voice-pipeline';
+export { VoicePipeline, createVoicePipeline } from './voice-pipeline.js';
 // Types
-export * from './types';
+export * from './types.js';
 // Backends
-export * from './backends';
+export * from './backends/index.js';
 // Services
-export * from './services';
+export * from './services/index.js';
 //# sourceMappingURL=index.js.map

package/dist/server/handler.js

@@ -5,8 +5,8 @@
  * Supports capability negotiation - skips STT/TTS when client handles them.
  * Each session has its own conversation history, backends, and state.
  */
-import { generateId } from '../client/protocol';
-import { float32ToBase64Node, base64ToFloat32Node, concatFloat32Arrays } from './encoding';
+import { generateId } from '../client/protocol.js';
+import { float32ToBase64Node, base64ToFloat32Node, concatFloat32Arrays } from './encoding.js';
 const DEFAULT_SILENCE_TIMEOUT_MS = 5000;
 const DEFAULT_MAX_RECORDING_MS = 90000;
 /**

package/dist/server/index.js

@@ -24,9 +24,9 @@
  * });
  * ```
  */
-export { createPipelineHandler, PipelineHandler, PipelineSession } from './handler';
+export { createPipelineHandler, PipelineHandler, PipelineSession } from './handler.js';
 // Re-export protocol types for server use
-export { float32ToBase64, base64ToFloat32, generateId, } from '../client/protocol';
+export { float32ToBase64, base64ToFloat32, generateId, } from '../client/protocol.js';
 // Server-side encoding utilities (use Buffer for efficiency in Node.js)
-export { encodeWav, float32ToBase64Node, base64ToFloat32Node } from './encoding';
+export { encodeWav, float32ToBase64Node, base64ToFloat32Node } from './encoding.js';
 //# sourceMappingURL=index.js.map

package/dist/services/index.js

@@ -1,3 +1,3 @@
-export { TextNormalizer } from './text-normalizer';
-export { LLMLogger, LLMConversationTracker, getDefaultLogger, getDefaultTracker } from './llm-logger';
+export { TextNormalizer } from './text-normalizer.js';
+export { LLMLogger, LLMConversationTracker, getDefaultLogger, getDefaultTracker } from './llm-logger.js';
 //# sourceMappingURL=index.js.map

package/dist/setup.js

@@ -19,9 +19,9 @@ import { spawn, execSync, spawnSync } from 'child_process';
 import { existsSync, readFileSync, mkdirSync, unlinkSync, statSync, createReadStream } from 'fs';
 import { createHash } from 'crypto';
 import { basename, join } from 'path';
-import { getModelsDir } from './cache';
+import { getModelsDir } from './cache.js';
 // Re-export cache status utilities so consumers only need one import
-export { checkModelsInstalled, getCacheStatus, getCacheDir, getModelsDir, getBinDir } from './cache';
+export { checkModelsInstalled, getCacheStatus, getCacheDir, getModelsDir, getBinDir } from './cache.js';
 // ============ Internal Helpers ============
 function formatBytes(bytes) {
     if (bytes < 1024)

package/dist/voice-pipeline.js

@@ -7,7 +7,7 @@
  *
  * The pipeline is stateless - callers manage conversation history via ConversationContext.
  */
-import { TextNormalizer, getDefaultLogger } from './services';
+import { TextNormalizer, getDefaultLogger } from './services/index.js';
 /** Maximum number of tool call iterations to prevent infinite loops */
 const MAX_TOOL_ITERATIONS = 10;
 /** Default filler phrases while executing tools */

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "modular-voice-agent-sdk",
-  "version": "2.9.1",
+  "version": "2.9.2",
   "description": "Modular isomorphic STT → LLM → TTS pipeline library for AI voice agents and voice assistants",
   "type": "module",
   "main": "./dist/index.js",
@@ -41,12 +41,13 @@
   },
   "files": [
     "dist",
-    "scripts",
+    "scripts/setup-binaries.sh",
+    "scripts/setup.sh",
     "README.md",
     "USAGE.md"
   ],
   "scripts": {
-    "build": "tsc -p tsconfig.build.json",
+    "build": "tsc -p tsconfig.build.json && node scripts/fix-esm-imports.mjs",
     "dev": "tsc -p tsconfig.build.json --watch",
     "prepublishOnly": "npm run build"
   },