@minded-ai/mindedjs 1.0.57 → 1.0.59

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (110) hide show
  1. package/dist/agent.d.ts +21 -11
  2. package/dist/agent.d.ts.map +1 -1
  3. package/dist/agent.js +87 -12
  4. package/dist/agent.js.map +1 -1
  5. package/dist/checkpointer/checkpointSaverFactory.d.ts.map +1 -1
  6. package/dist/checkpointer/checkpointSaverFactory.js +3 -2
  7. package/dist/checkpointer/checkpointSaverFactory.js.map +1 -1
  8. package/dist/cli/index.js +6 -5
  9. package/dist/cli/index.js.map +1 -1
  10. package/dist/edges/createDirectEdge.d.ts.map +1 -1
  11. package/dist/edges/createDirectEdge.js +2 -1
  12. package/dist/edges/createDirectEdge.js.map +1 -1
  13. package/dist/edges/createLogicalRouter.d.ts.map +1 -1
  14. package/dist/edges/createLogicalRouter.js +2 -1
  15. package/dist/edges/createLogicalRouter.js.map +1 -1
  16. package/dist/edges/createPromptRouter.d.ts.map +1 -1
  17. package/dist/edges/createPromptRouter.js +3 -2
  18. package/dist/edges/createPromptRouter.js.map +1 -1
  19. package/dist/index.d.ts +5 -2
  20. package/dist/index.d.ts.map +1 -1
  21. package/dist/index.js +3 -1
  22. package/dist/index.js.map +1 -1
  23. package/dist/nodes/addAppToolNode.d.ts.map +1 -1
  24. package/dist/nodes/addAppToolNode.js +23 -10
  25. package/dist/nodes/addAppToolNode.js.map +1 -1
  26. package/dist/nodes/addHumanInTheLoopNode.d.ts.map +1 -1
  27. package/dist/nodes/addHumanInTheLoopNode.js +2 -1
  28. package/dist/nodes/addHumanInTheLoopNode.js.map +1 -1
  29. package/dist/nodes/addPromptNode.d.ts.map +1 -1
  30. package/dist/nodes/addPromptNode.js +7 -2
  31. package/dist/nodes/addPromptNode.js.map +1 -1
  32. package/dist/nodes/addToolNode.d.ts.map +1 -1
  33. package/dist/nodes/addToolNode.js +2 -1
  34. package/dist/nodes/addToolNode.js.map +1 -1
  35. package/dist/nodes/addToolRunNode.d.ts.map +1 -1
  36. package/dist/nodes/addToolRunNode.js +2 -1
  37. package/dist/nodes/addToolRunNode.js.map +1 -1
  38. package/dist/nodes/addTriggerNode.d.ts.map +1 -1
  39. package/dist/nodes/addTriggerNode.js +2 -1
  40. package/dist/nodes/addTriggerNode.js.map +1 -1
  41. package/dist/platform/config.d.ts +3 -0
  42. package/dist/platform/config.d.ts.map +1 -1
  43. package/dist/platform/config.js +18 -1
  44. package/dist/platform/config.js.map +1 -1
  45. package/dist/platform/mindedCheckpointSaver.d.ts.map +1 -1
  46. package/dist/platform/mindedCheckpointSaver.js +3 -2
  47. package/dist/platform/mindedCheckpointSaver.js.map +1 -1
  48. package/dist/platform/mindedConnection.d.ts +4 -4
  49. package/dist/platform/mindedConnection.d.ts.map +1 -1
  50. package/dist/platform/mindedConnection.js +14 -10
  51. package/dist/platform/mindedConnection.js.map +1 -1
  52. package/dist/platform/mindedConnectionTypes.d.ts +48 -3
  53. package/dist/platform/mindedConnectionTypes.d.ts.map +1 -1
  54. package/dist/platform/mindedConnectionTypes.js +11 -3
  55. package/dist/platform/mindedConnectionTypes.js.map +1 -1
  56. package/dist/platform/piiGateway/gateway.d.ts.map +1 -1
  57. package/dist/platform/piiGateway/gateway.js +10 -2
  58. package/dist/platform/piiGateway/gateway.js.map +1 -1
  59. package/dist/types/Agent.types.d.ts +13 -0
  60. package/dist/types/Agent.types.d.ts.map +1 -1
  61. package/dist/types/Agent.types.js.map +1 -1
  62. package/dist/types/Flows.types.d.ts +15 -2
  63. package/dist/types/Flows.types.d.ts.map +1 -1
  64. package/dist/types/Flows.types.js +1 -0
  65. package/dist/types/Flows.types.js.map +1 -1
  66. package/dist/types/LangGraph.types.d.ts.map +1 -1
  67. package/dist/types/LangGraph.types.js +16 -1
  68. package/dist/types/LangGraph.types.js.map +1 -1
  69. package/dist/types/Voice.types.d.ts +5 -0
  70. package/dist/types/Voice.types.d.ts.map +1 -0
  71. package/dist/types/Voice.types.js +3 -0
  72. package/dist/types/Voice.types.js.map +1 -0
  73. package/dist/utils/logger.d.ts +2 -0
  74. package/dist/utils/logger.d.ts.map +1 -0
  75. package/dist/utils/logger.js +28 -0
  76. package/dist/utils/logger.js.map +1 -0
  77. package/dist/voice/elevenLabsUtils.d.ts +65 -0
  78. package/dist/voice/elevenLabsUtils.d.ts.map +1 -0
  79. package/dist/voice/elevenLabsUtils.js +4 -0
  80. package/dist/voice/elevenLabsUtils.js.map +1 -0
  81. package/dist/voice/voiceSession.d.ts +50 -0
  82. package/dist/voice/voiceSession.d.ts.map +1 -0
  83. package/dist/voice/voiceSession.js +244 -0
  84. package/dist/voice/voiceSession.js.map +1 -0
  85. package/package.json +7 -2
  86. package/src/agent.ts +103 -24
  87. package/src/checkpointer/checkpointSaverFactory.ts +3 -2
  88. package/src/cli/index.ts +7 -9
  89. package/src/edges/createDirectEdge.ts +2 -1
  90. package/src/edges/createLogicalRouter.ts +2 -1
  91. package/src/edges/createPromptRouter.ts +3 -2
  92. package/src/index.ts +6 -1
  93. package/src/nodes/addAppToolNode.ts +23 -10
  94. package/src/nodes/addHumanInTheLoopNode.ts +2 -1
  95. package/src/nodes/addPromptNode.ts +6 -2
  96. package/src/nodes/addToolNode.ts +2 -1
  97. package/src/nodes/addToolRunNode.ts +2 -1
  98. package/src/nodes/addTriggerNode.ts +7 -6
  99. package/src/platform/config.ts +21 -1
  100. package/src/platform/mindedCheckpointSaver.ts +3 -2
  101. package/src/platform/mindedConnection.ts +21 -14
  102. package/src/platform/mindedConnectionTypes.ts +56 -3
  103. package/src/platform/piiGateway/gateway.ts +10 -2
  104. package/src/types/Agent.types.ts +14 -0
  105. package/src/types/Flows.types.ts +15 -1
  106. package/src/types/LangGraph.types.ts +15 -1
  107. package/src/types/Voice.types.ts +4 -0
  108. package/src/utils/logger.ts +22 -0
  109. package/src/voice/elevenLabsUtils.ts +81 -0
  110. package/src/voice/voiceSession.ts +277 -0
@@ -0,0 +1,81 @@
1
+ // ----------------- ElevenLabs WebSocket event types -----------------
2
+
3
+ interface BaseEvent {
4
+ type: string;
5
+ }
6
+
7
+ interface UserTranscriptEvent extends BaseEvent {
8
+ type: 'user_transcript';
9
+ user_transcription_event: {
10
+ user_transcript: string;
11
+ };
12
+ }
13
+
14
+ interface AgentResponseEvent extends BaseEvent {
15
+ type: 'agent_response';
16
+ agent_response_event: {
17
+ agent_response: string;
18
+ };
19
+ }
20
+
21
+ interface AudioResponseEvent extends BaseEvent {
22
+ type: 'audio';
23
+ audio_event: {
24
+ audio_base_64: string;
25
+ event_id: number;
26
+ };
27
+ }
28
+
29
+ interface InterruptionEvent extends BaseEvent {
30
+ type: 'interruption';
31
+ interruption_event: {
32
+ event_id: number;
33
+ };
34
+ }
35
+
36
+ interface PingEvent extends BaseEvent {
37
+ type: 'ping';
38
+ ping_event: {
39
+ event_id: number;
40
+ ping_ms: number;
41
+ };
42
+ }
43
+
44
+ interface ConversationInitiationMetadataEvent extends BaseEvent {
45
+ type: 'conversation_initiation_metadata';
46
+ conversation_initiation_metadata_event: {
47
+ conversation_id: string;
48
+ };
49
+ }
50
+
51
+ export interface AgentResponseCorrectionEvent extends BaseEvent {
52
+ type: 'agent_response_correction';
53
+ agent_response_correction_event: {
54
+ corrected_agent_response: string;
55
+ original_agent_response: string;
56
+ };
57
+ }
58
+
59
+ /** Union of all possible WebSocket events we care about */
60
+ export type ElevenLabsWebSocketEvent =
61
+ | UserTranscriptEvent
62
+ | AgentResponseEvent
63
+ | AudioResponseEvent
64
+ | InterruptionEvent
65
+ | PingEvent
66
+ | ConversationInitiationMetadataEvent
67
+ | AgentResponseCorrectionEvent;
68
+
69
+ /** Shape of the message sent to ElevenLabs when starting a conversation */
70
+ export interface ConversationInitiationClientData {
71
+ type: 'conversation_initiation_client_data';
72
+ conversation_config_override: {
73
+ agent: {
74
+ first_message: string;
75
+ language: string;
76
+ };
77
+ tts?: {
78
+ voice_id: string;
79
+ };
80
+ };
81
+ }
@@ -0,0 +1,277 @@
1
+ import { WebSocket } from 'ws';
2
+ import { Agent } from '../agent';
3
+ import {
4
+ MindedConnectionSocketMessageType,
5
+ } from '../platform/mindedConnectionTypes';
6
+ import { getConfig } from '../platform/config';
7
+ import {
8
+ ConversationInitiationClientData,
9
+ ElevenLabsWebSocketEvent,
10
+ } from './elevenLabsUtils';
11
+ import { AIMessage, BaseMessage, HumanMessage } from '@langchain/core/messages';
12
+ import { v4 as uuidv4 } from 'uuid';
13
+ import { logger } from '../utils/logger';
14
+
15
+ /**
16
+ * Voice Conversation class for managing individual ElevenLabs voice conversations
17
+ */
18
+ export class VoiceSession {
19
+ private agent: Agent;
20
+ private sessionId: string;
21
+ private firstMessage: string;
22
+ private voiceId?: string;
23
+ private elevenLabsSocket?: WebSocket;
24
+
25
+ private onAudioCallback?: (data: string) => void;
26
+ private onInterruptionCallback?: () => void;
27
+ private onMessageCallback?: (text: string, message: BaseMessage) => void;
28
+ private onDisconnectCallback?: () => void;
29
+
30
+ constructor({ agent, sessionId, firstMessage, voiceId }: { agent: Agent; sessionId: string; firstMessage: string; voiceId?: string }) {
31
+ logger.debug('Starting voice session', { sessionId, firstMessage, voiceId });
32
+ this.agent = agent;
33
+ this.sessionId = sessionId;
34
+ this.firstMessage = firstMessage;
35
+ this.voiceId = voiceId;
36
+ }
37
+
38
+ /**
39
+ * Initialize the voice conversation connection
40
+ */
41
+ public async init(): Promise<void> {
42
+ const { elevenLabsKey, elevenLabsAgentId } = getConfig();
43
+
44
+ if (!elevenLabsKey) {
45
+ throw new Error('Missing ElevenLabs key - set ELEVEN_LABS_KEY env var');
46
+ }
47
+
48
+ if (!elevenLabsAgentId) {
49
+ throw new Error('Missing ElevenLabs agent id - set ELEVEN_LABS_AGENT_ID env var');
50
+ }
51
+
52
+ const signedUrl = await this.getSignedUrl({
53
+ agentId: elevenLabsAgentId,
54
+ apiKey: elevenLabsKey
55
+ });
56
+
57
+ this.elevenLabsSocket = new WebSocket(signedUrl);
58
+ this.setupSocketHandlers();
59
+ }
60
+
61
+ private async sendToElevenLabs(message: string): Promise<void> {
62
+ if (!this.elevenLabsSocket) {
63
+ throw new Error('Socket not initialized');
64
+ }
65
+ await this.waitForSocketOpen(this.elevenLabsSocket);
66
+ this.elevenLabsSocket.send(message);
67
+ }
68
+
69
+ private async waitForSocketOpen(socket: WebSocket): Promise<void> {
70
+ if (socket.readyState === WebSocket.OPEN) {
71
+ return;
72
+ }
73
+ return new Promise((resolve, reject) => {
74
+ const timeoutMs = 10000;
75
+ const intervalMs = 100;
76
+ let tries = 0;
77
+ const interval = setInterval(() => {
78
+ if (socket.readyState === WebSocket.OPEN) {
79
+ clearInterval(interval);
80
+ resolve();
81
+ }
82
+ tries++;
83
+ if (tries >= timeoutMs) {
84
+ clearInterval(interval);
85
+ reject(new Error('Socket not open'));
86
+ }
87
+ }, intervalMs);
88
+ });
89
+ }
90
+
91
+ private setupSocketHandlers(): void {
92
+ const socket = this.elevenLabsSocket!; // non-null assertion once, we ensured it's assigned in init()
93
+
94
+ socket.onopen = () => {
95
+ logger.debug('Connected to ElevenLabs');
96
+ const initiationData: ConversationInitiationClientData = {
97
+ type: 'conversation_initiation_client_data',
98
+ conversation_config_override: {
99
+ agent: {
100
+ first_message: this.firstMessage,
101
+ language: 'en',
102
+ },
103
+ ...(this.voiceId
104
+ ? {
105
+ tts: {
106
+ voice_id: this.voiceId,
107
+ },
108
+ }
109
+ : {}),
110
+ } as ConversationInitiationClientData['conversation_config_override'],
111
+ };
112
+
113
+ this.sendToElevenLabs(JSON.stringify(initiationData));
114
+ this.sendToElevenLabs(JSON.stringify({ type: 'contextual_update', text: JSON.stringify({ sessionId: this.sessionId, mindedToken: getConfig().token! }) }));
115
+ this.addFirstMessageToState();
116
+ };
117
+
118
+ socket.onclose = () => {
119
+ logger.debug('Disconnected from ElevenLabs');
120
+ this.onDisconnectCallback?.();
121
+ this.agent.voiceSessions.delete(this.sessionId);
122
+ };
123
+
124
+ socket.onerror = (err: unknown) => {
125
+ logger.error('[ElevenLabsVoice] socket error', err);
126
+ };
127
+
128
+ socket.onmessage = async (event: any) => {
129
+ const data: ElevenLabsWebSocketEvent = JSON.parse(event.data.toString());
130
+
131
+ switch (data.type) {
132
+ case 'ping':
133
+ setTimeout(() => {
134
+ this.sendToElevenLabs(JSON.stringify({ type: 'pong', event_id: data.ping_event.event_id }));
135
+ }, data.ping_event.ping_ms);
136
+ break;
137
+ case 'user_transcript':
138
+ logger.debug('User transcript received:', data.user_transcription_event.user_transcript);
139
+ if (this.onMessageCallback) {
140
+ this.onMessageCallback(data.user_transcription_event.user_transcript, new HumanMessage({ content: data.user_transcription_event.user_transcript }));
141
+ }
142
+ break;
143
+ case 'agent_response':
144
+ if (this.onMessageCallback) {
145
+ this.onMessageCallback(data.agent_response_event.agent_response, new AIMessage({ content: data.agent_response_event.agent_response }));
146
+ }
147
+ break;
148
+ case 'interruption':
149
+ logger.debug('Interruption received');
150
+ this.onInterruptionCallback?.();
151
+ // Send interruption event to dashboard if connected
152
+ if (getConfig().dashboardConnected) {
153
+ try {
154
+ this.agent.mindedConnection?.emit(MindedConnectionSocketMessageType.DASHBOARD_VOICE_INTERRUPTION, {
155
+ type: MindedConnectionSocketMessageType.DASHBOARD_VOICE_INTERRUPTION,
156
+ sessionId: this.sessionId,
157
+ timestamp: Date.now(),
158
+ });
159
+ } catch (error) {
160
+ logger.error('[ElevenLabsVoice] Error sending interruption to dashboard', error);
161
+ }
162
+ }
163
+ break;
164
+ case 'audio':
165
+ if (this.onAudioCallback) {
166
+ this.onAudioCallback(data.audio_event.audio_base_64);
167
+ }
168
+ if (getConfig().dashboardConnected) {
169
+ try {
170
+ this.agent.mindedConnection?.emit(MindedConnectionSocketMessageType.DASHBOARD_VOICE_AGENT_AUDIO, {
171
+ sessionId: this.sessionId,
172
+ audioData: data.audio_event.audio_base_64,
173
+ type: MindedConnectionSocketMessageType.DASHBOARD_VOICE_AGENT_AUDIO,
174
+ timestamp: Date.now(),
175
+ });
176
+ } catch (error) {
177
+ logger.error('[ElevenLabsVoice] Error sending audio to dashboard', error);
178
+ }
179
+ }
180
+ break;
181
+ case 'conversation_initiation_metadata':
182
+ logger.debug('ElevenLabs conversation initiation metadata', data);
183
+ break;
184
+ case 'agent_response_correction':
185
+ try {
186
+ logger.debug('Agent response correction received', data.agent_response_correction_event);
187
+ await this.updateAgentResponse(data.agent_response_correction_event.original_agent_response, data.agent_response_correction_event.corrected_agent_response);
188
+ } catch (error) {
189
+ logger.error('[ElevenLabsVoice] Error updating agent response', error);
190
+ }
191
+ break;
192
+ default:
193
+ logger.debug('Received unknown message from ElevenLabs', data);
194
+ break;
195
+ }
196
+ };
197
+ }
198
+
199
+ private async addFirstMessageToState(): Promise<void> {
200
+ const graphState = await this.agent.compiledGraph.getState(this.agent.getLangraphConfig(this.sessionId));
201
+ if (graphState.values.messages.length === 0) {
202
+ logger.info('Adding first message to state', this.firstMessage);
203
+ await this.agent.compiledGraph.updateState(this.agent.getLangraphConfig(this.sessionId), {
204
+ messages: [new AIMessage({ content: this.firstMessage, id: uuidv4() })],
205
+ });
206
+ }
207
+ }
208
+
209
+ private async updateAgentResponse(originalAgentResponse: string, correctedAgentResponse: string): Promise<void> {
210
+ const graphState = await this.agent.compiledGraph.getState(this.agent.getLangraphConfig(this.sessionId));
211
+ const agentMessage: AIMessage | null = graphState.values.messages.find((message: BaseMessage) => message.content === originalAgentResponse && message instanceof AIMessage);
212
+ if (agentMessage) {
213
+ agentMessage.content = correctedAgentResponse;
214
+ agentMessage.additional_kwargs = agentMessage.additional_kwargs || {};
215
+ agentMessage.additional_kwargs.update = true;
216
+ const currentNodeId = graphState.tasks[graphState.tasks.length - 1].name;
217
+ await this.agent.compiledGraph.updateState(this.agent.getLangraphConfig(this.sessionId), {
218
+ messages: [agentMessage],
219
+ }, currentNodeId);
220
+ }
221
+ else {
222
+ logger.warn('Agent message not found for correction', originalAgentResponse);
223
+ }
224
+ }
225
+
226
+ public hangup(): void {
227
+ logger.info('Hanging up voice session', this.sessionId);
228
+ this.onDisconnectCallback?.();
229
+ this.elevenLabsSocket?.close();
230
+ }
231
+
232
+ /**
233
+ * Set callback for audio data in base64 format
234
+ */
235
+ onAudio(callback: (data: string) => void): void {
236
+ this.onAudioCallback = callback;
237
+ }
238
+
239
+ /**
240
+ * Set callback for interruption events
241
+ */
242
+ onInterruption(callback: () => void): void {
243
+ this.onInterruptionCallback = callback;
244
+ }
245
+
246
+ /**
247
+ * Set callback for disconnect events
248
+ */
249
+ onDisconnect(callback: () => void): void {
250
+ this.onDisconnectCallback = callback;
251
+ }
252
+
253
+ /**
254
+ * User audio in base64 format
255
+ */
256
+ sendAudio(audioData: string): void {
257
+ this.sendToElevenLabs(JSON.stringify({ user_audio_chunk: audioData }));
258
+ }
259
+
260
+ /** Fetch signed URL from ElevenLabs API */
261
+ private async getSignedUrl({ agentId, apiKey }: { agentId: string; apiKey: string }): Promise<string> {
262
+ const response = await fetch(
263
+ `https://api.elevenlabs.io/v1/convai/conversation/get_signed_url?agent_id=${agentId}`,
264
+ {
265
+ method: 'GET',
266
+ headers: {
267
+ 'xi-api-key': apiKey,
268
+ },
269
+ },
270
+ );
271
+ if (!response.ok) {
272
+ throw new Error(`Failed to fetch signed url - status ${response.status}`);
273
+ }
274
+ const body = await response.json();
275
+ return body.signed_url;
276
+ }
277
+ }