retell-sdk 4.42.0 → 4.44.0

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.
@@ -103,15 +103,19 @@ export interface LlmResponse {
103
103
  * Whether the Retell LLM Response Engine is published.
104
104
  */
105
105
  is_published?: boolean;
106
+ /**
107
+ * Knowledge base configuration for RAG retrieval.
108
+ */
109
+ kb_config?: LlmResponse.KBConfig | null;
106
110
  /**
107
111
  * A list of knowledge base ids to use for this resource. Set to null to remove all
108
112
  * knowledge bases.
109
113
  */
110
114
  knowledge_base_ids?: Array<string> | null;
111
115
  /**
112
- * Select the underlying text LLM. If not set, would default to gpt-4o.
116
+ * Select the underlying text LLM. If not set, would default to gpt-4.1.
113
117
  */
114
- model?: 'gpt-4o' | 'gpt-4o-mini' | 'gpt-4.1' | 'gpt-4.1-mini' | 'gpt-4.1-nano' | 'claude-3.7-sonnet' | 'claude-3.5-haiku' | 'gemini-2.0-flash' | 'gemini-2.0-flash-lite' | null;
118
+ model?: 'gpt-5' | 'gpt-5-mini' | 'gpt-5-nano' | 'gpt-4o' | 'gpt-4o-mini' | 'gpt-4.1' | 'gpt-4.1-mini' | 'gpt-4.1-nano' | 'claude-3.7-sonnet' | 'claude-3.5-haiku' | 'gemini-2.0-flash' | 'gemini-2.0-flash-lite' | 'gemini-2.5-flash' | 'gemini-2.5-flash-lite' | null;
115
119
  /**
116
120
  * If set to true, will enable fast tier, which uses high priority pool with more
117
121
  * dedicated resource to ensure lower and more consistent latency, default to
@@ -129,7 +133,7 @@ export interface LlmResponse {
129
133
  * Select the underlying speech to speech model. Can only set this or model, not
130
134
  * both.
131
135
  */
132
- s2s_model?: 'gpt-4o-realtime' | 'gpt-4o-mini-realtime' | null;
136
+ s2s_model?: 'gpt-4o-realtime' | 'gpt-4o-mini-realtime' | 'gpt-realtime' | null;
133
137
  /**
134
138
  * Name of the starting state. Required if states is not empty.
135
139
  */
@@ -202,6 +206,11 @@ export declare namespace LlmResponse {
202
206
  * The type of transfer destination.
203
207
  */
204
208
  type: 'predefined';
209
+ /**
210
+ * Extension digits to dial after the main number connects. Sent via DTMF. Allow
211
+ * digits, '\*', '#'.
212
+ */
213
+ extension?: string;
205
214
  }
206
215
  interface TransferDestinationInferred {
207
216
  /**
@@ -607,15 +616,36 @@ export declare namespace LlmResponse {
607
616
  name: string;
608
617
  type: 'mcp';
609
618
  /**
610
- * The input schema of the MCP tool.
619
+ * The description for the sentence agent say during execution. Only applicable
620
+ * when speak_during_execution is true. Can write what to say or even provide
621
+ * examples. The default is "The message you will say to callee when calling this
622
+ * tool. Make sure it fits into the conversation smoothly.".
611
623
  */
612
- input_schema?: {
613
- [key: string]: string;
614
- };
624
+ execution_message_description?: string;
615
625
  /**
616
626
  * Unique id of the MCP.
617
627
  */
618
628
  mcp_id?: string;
629
+ /**
630
+ * Response variables to add to dynamic variables, key is the variable name, value
631
+ * is the path to the variable in the response
632
+ */
633
+ response_variables?: {
634
+ [key: string]: string;
635
+ };
636
+ /**
637
+ * Determines whether the agent would call LLM another time and speak when the
638
+ * result of function is obtained. Usually this needs to get turned on so user can
639
+ * get update for the function call.
640
+ */
641
+ speak_after_execution?: boolean;
642
+ /**
643
+ * Determines whether the agent would say sentence like "One moment, let me check
644
+ * that." when executing the function. Recommend to turn on if your function call
645
+ * takes over 1s (including network) to complete, so that your agent remains
646
+ * responsive.
647
+ */
648
+ speak_during_execution?: boolean;
619
649
  }
620
650
  interface SendSMSTool {
621
651
  /**
@@ -649,6 +679,19 @@ export declare namespace LlmResponse {
649
679
  type?: 'inferred';
650
680
  }
651
681
  }
682
+ /**
683
+ * Knowledge base configuration for RAG retrieval.
684
+ */
685
+ interface KBConfig {
686
+ /**
687
+ * Similarity threshold for filtering search results
688
+ */
689
+ filter_score?: number;
690
+ /**
691
+ * Max number of knowledge base chunks to retrieve
692
+ */
693
+ top_k?: number;
694
+ }
652
695
  interface State {
653
696
  /**
654
697
  * Name of the state, must be unique for each state. Must be consisted of a-z, A-Z,
@@ -776,6 +819,11 @@ export declare namespace LlmResponse {
776
819
  * The type of transfer destination.
777
820
  */
778
821
  type: 'predefined';
822
+ /**
823
+ * Extension digits to dial after the main number connects. Sent via DTMF. Allow
824
+ * digits, '\*', '#'.
825
+ */
826
+ extension?: string;
779
827
  }
780
828
  interface TransferDestinationInferred {
781
829
  /**
@@ -1181,15 +1229,36 @@ export declare namespace LlmResponse {
1181
1229
  name: string;
1182
1230
  type: 'mcp';
1183
1231
  /**
1184
- * The input schema of the MCP tool.
1232
+ * The description for the sentence agent say during execution. Only applicable
1233
+ * when speak_during_execution is true. Can write what to say or even provide
1234
+ * examples. The default is "The message you will say to callee when calling this
1235
+ * tool. Make sure it fits into the conversation smoothly.".
1185
1236
  */
1186
- input_schema?: {
1187
- [key: string]: string;
1188
- };
1237
+ execution_message_description?: string;
1189
1238
  /**
1190
1239
  * Unique id of the MCP.
1191
1240
  */
1192
1241
  mcp_id?: string;
1242
+ /**
1243
+ * Response variables to add to dynamic variables, key is the variable name, value
1244
+ * is the path to the variable in the response
1245
+ */
1246
+ response_variables?: {
1247
+ [key: string]: string;
1248
+ };
1249
+ /**
1250
+ * Determines whether the agent would call LLM another time and speak when the
1251
+ * result of function is obtained. Usually this needs to get turned on so user can
1252
+ * get update for the function call.
1253
+ */
1254
+ speak_after_execution?: boolean;
1255
+ /**
1256
+ * Determines whether the agent would say sentence like "One moment, let me check
1257
+ * that." when executing the function. Recommend to turn on if your function call
1258
+ * takes over 1s (including network) to complete, so that your agent remains
1259
+ * responsive.
1260
+ */
1261
+ speak_during_execution?: boolean;
1193
1262
  }
1194
1263
  interface SendSMSTool {
1195
1264
  /**
@@ -1258,15 +1327,19 @@ export interface LlmCreateParams {
1258
1327
  * - Tools of LLM (no state) = general tools
1259
1328
  */
1260
1329
  general_tools?: Array<LlmCreateParams.EndCallTool | LlmCreateParams.TransferCallTool | LlmCreateParams.CheckAvailabilityCalTool | LlmCreateParams.BookAppointmentCalTool | LlmCreateParams.PressDigitTool | LlmCreateParams.CustomTool | LlmCreateParams.ExtractDynamicVariableTool | LlmCreateParams.AgentSwapTool | LlmCreateParams.McpTool | LlmCreateParams.SendSMSTool> | null;
1330
+ /**
1331
+ * Knowledge base configuration for RAG retrieval.
1332
+ */
1333
+ kb_config?: LlmCreateParams.KBConfig | null;
1261
1334
  /**
1262
1335
  * A list of knowledge base ids to use for this resource. Set to null to remove all
1263
1336
  * knowledge bases.
1264
1337
  */
1265
1338
  knowledge_base_ids?: Array<string> | null;
1266
1339
  /**
1267
- * Select the underlying text LLM. If not set, would default to gpt-4o.
1340
+ * Select the underlying text LLM. If not set, would default to gpt-4.1.
1268
1341
  */
1269
- model?: 'gpt-4o' | 'gpt-4o-mini' | 'gpt-4.1' | 'gpt-4.1-mini' | 'gpt-4.1-nano' | 'claude-3.7-sonnet' | 'claude-3.5-haiku' | 'gemini-2.0-flash' | 'gemini-2.0-flash-lite' | null;
1342
+ model?: 'gpt-5' | 'gpt-5-mini' | 'gpt-5-nano' | 'gpt-4o' | 'gpt-4o-mini' | 'gpt-4.1' | 'gpt-4.1-mini' | 'gpt-4.1-nano' | 'claude-3.7-sonnet' | 'claude-3.5-haiku' | 'gemini-2.0-flash' | 'gemini-2.0-flash-lite' | 'gemini-2.5-flash' | 'gemini-2.5-flash-lite' | null;
1270
1343
  /**
1271
1344
  * If set to true, will enable fast tier, which uses high priority pool with more
1272
1345
  * dedicated resource to ensure lower and more consistent latency, default to
@@ -1284,7 +1357,7 @@ export interface LlmCreateParams {
1284
1357
  * Select the underlying speech to speech model. Can only set this or model, not
1285
1358
  * both.
1286
1359
  */
1287
- s2s_model?: 'gpt-4o-realtime' | 'gpt-4o-mini-realtime' | null;
1360
+ s2s_model?: 'gpt-4o-realtime' | 'gpt-4o-mini-realtime' | 'gpt-realtime' | null;
1288
1361
  /**
1289
1362
  * Name of the starting state. Required if states is not empty.
1290
1363
  */
@@ -1357,6 +1430,11 @@ export declare namespace LlmCreateParams {
1357
1430
  * The type of transfer destination.
1358
1431
  */
1359
1432
  type: 'predefined';
1433
+ /**
1434
+ * Extension digits to dial after the main number connects. Sent via DTMF. Allow
1435
+ * digits, '\*', '#'.
1436
+ */
1437
+ extension?: string;
1360
1438
  }
1361
1439
  interface TransferDestinationInferred {
1362
1440
  /**
@@ -1762,15 +1840,36 @@ export declare namespace LlmCreateParams {
1762
1840
  name: string;
1763
1841
  type: 'mcp';
1764
1842
  /**
1765
- * The input schema of the MCP tool.
1843
+ * The description for the sentence agent say during execution. Only applicable
1844
+ * when speak_during_execution is true. Can write what to say or even provide
1845
+ * examples. The default is "The message you will say to callee when calling this
1846
+ * tool. Make sure it fits into the conversation smoothly.".
1847
+ */
1848
+ execution_message_description?: string;
1849
+ /**
1850
+ * Unique id of the MCP.
1851
+ */
1852
+ mcp_id?: string;
1853
+ /**
1854
+ * Response variables to add to dynamic variables, key is the variable name, value
1855
+ * is the path to the variable in the response
1766
1856
  */
1767
- input_schema?: {
1857
+ response_variables?: {
1768
1858
  [key: string]: string;
1769
1859
  };
1770
1860
  /**
1771
- * Unique id of the MCP.
1861
+ * Determines whether the agent would call LLM another time and speak when the
1862
+ * result of function is obtained. Usually this needs to get turned on so user can
1863
+ * get update for the function call.
1772
1864
  */
1773
- mcp_id?: string;
1865
+ speak_after_execution?: boolean;
1866
+ /**
1867
+ * Determines whether the agent would say sentence like "One moment, let me check
1868
+ * that." when executing the function. Recommend to turn on if your function call
1869
+ * takes over 1s (including network) to complete, so that your agent remains
1870
+ * responsive.
1871
+ */
1872
+ speak_during_execution?: boolean;
1774
1873
  }
1775
1874
  interface SendSMSTool {
1776
1875
  /**
@@ -1804,6 +1903,19 @@ export declare namespace LlmCreateParams {
1804
1903
  type?: 'inferred';
1805
1904
  }
1806
1905
  }
1906
+ /**
1907
+ * Knowledge base configuration for RAG retrieval.
1908
+ */
1909
+ interface KBConfig {
1910
+ /**
1911
+ * Similarity threshold for filtering search results
1912
+ */
1913
+ filter_score?: number;
1914
+ /**
1915
+ * Max number of knowledge base chunks to retrieve
1916
+ */
1917
+ top_k?: number;
1918
+ }
1807
1919
  interface State {
1808
1920
  /**
1809
1921
  * Name of the state, must be unique for each state. Must be consisted of a-z, A-Z,
@@ -1931,6 +2043,11 @@ export declare namespace LlmCreateParams {
1931
2043
  * The type of transfer destination.
1932
2044
  */
1933
2045
  type: 'predefined';
2046
+ /**
2047
+ * Extension digits to dial after the main number connects. Sent via DTMF. Allow
2048
+ * digits, '\*', '#'.
2049
+ */
2050
+ extension?: string;
1934
2051
  }
1935
2052
  interface TransferDestinationInferred {
1936
2053
  /**
@@ -2336,15 +2453,36 @@ export declare namespace LlmCreateParams {
2336
2453
  name: string;
2337
2454
  type: 'mcp';
2338
2455
  /**
2339
- * The input schema of the MCP tool.
2456
+ * The description for the sentence agent say during execution. Only applicable
2457
+ * when speak_during_execution is true. Can write what to say or even provide
2458
+ * examples. The default is "The message you will say to callee when calling this
2459
+ * tool. Make sure it fits into the conversation smoothly.".
2340
2460
  */
2341
- input_schema?: {
2342
- [key: string]: string;
2343
- };
2461
+ execution_message_description?: string;
2344
2462
  /**
2345
2463
  * Unique id of the MCP.
2346
2464
  */
2347
2465
  mcp_id?: string;
2466
+ /**
2467
+ * Response variables to add to dynamic variables, key is the variable name, value
2468
+ * is the path to the variable in the response
2469
+ */
2470
+ response_variables?: {
2471
+ [key: string]: string;
2472
+ };
2473
+ /**
2474
+ * Determines whether the agent would call LLM another time and speak when the
2475
+ * result of function is obtained. Usually this needs to get turned on so user can
2476
+ * get update for the function call.
2477
+ */
2478
+ speak_after_execution?: boolean;
2479
+ /**
2480
+ * Determines whether the agent would say sentence like "One moment, let me check
2481
+ * that." when executing the function. Recommend to turn on if your function call
2482
+ * takes over 1s (including network) to complete, so that your agent remains
2483
+ * responsive.
2484
+ */
2485
+ speak_during_execution?: boolean;
2348
2486
  }
2349
2487
  interface SendSMSTool {
2350
2488
  /**
@@ -2426,15 +2564,20 @@ export interface LlmUpdateParams {
2426
2564
  * - Tools of LLM (no state) = general tools
2427
2565
  */
2428
2566
  general_tools?: Array<LlmUpdateParams.EndCallTool | LlmUpdateParams.TransferCallTool | LlmUpdateParams.CheckAvailabilityCalTool | LlmUpdateParams.BookAppointmentCalTool | LlmUpdateParams.PressDigitTool | LlmUpdateParams.CustomTool | LlmUpdateParams.ExtractDynamicVariableTool | LlmUpdateParams.AgentSwapTool | LlmUpdateParams.McpTool | LlmUpdateParams.SendSMSTool> | null;
2567
+ /**
2568
+ * Body param: Knowledge base configuration for RAG retrieval.
2569
+ */
2570
+ kb_config?: LlmUpdateParams.KBConfig | null;
2429
2571
  /**
2430
2572
  * Body param: A list of knowledge base ids to use for this resource. Set to null
2431
2573
  * to remove all knowledge bases.
2432
2574
  */
2433
2575
  knowledge_base_ids?: Array<string> | null;
2434
2576
  /**
2435
- * Body param: Select the underlying text LLM. If not set, would default to gpt-4o.
2577
+ * Body param: Select the underlying text LLM. If not set, would default to
2578
+ * gpt-4.1.
2436
2579
  */
2437
- model?: 'gpt-4o' | 'gpt-4o-mini' | 'gpt-4.1' | 'gpt-4.1-mini' | 'gpt-4.1-nano' | 'claude-3.7-sonnet' | 'claude-3.5-haiku' | 'gemini-2.0-flash' | 'gemini-2.0-flash-lite' | null;
2580
+ model?: 'gpt-5' | 'gpt-5-mini' | 'gpt-5-nano' | 'gpt-4o' | 'gpt-4o-mini' | 'gpt-4.1' | 'gpt-4.1-mini' | 'gpt-4.1-nano' | 'claude-3.7-sonnet' | 'claude-3.5-haiku' | 'gemini-2.0-flash' | 'gemini-2.0-flash-lite' | 'gemini-2.5-flash' | 'gemini-2.5-flash-lite' | null;
2438
2581
  /**
2439
2582
  * Body param: If set to true, will enable fast tier, which uses high priority pool
2440
2583
  * with more dedicated resource to ensure lower and more consistent latency,
@@ -2452,7 +2595,7 @@ export interface LlmUpdateParams {
2452
2595
  * Body param: Select the underlying speech to speech model. Can only set this or
2453
2596
  * model, not both.
2454
2597
  */
2455
- s2s_model?: 'gpt-4o-realtime' | 'gpt-4o-mini-realtime' | null;
2598
+ s2s_model?: 'gpt-4o-realtime' | 'gpt-4o-mini-realtime' | 'gpt-realtime' | null;
2456
2599
  /**
2457
2600
  * Body param: Name of the starting state. Required if states is not empty.
2458
2601
  */
@@ -2525,6 +2668,11 @@ export declare namespace LlmUpdateParams {
2525
2668
  * The type of transfer destination.
2526
2669
  */
2527
2670
  type: 'predefined';
2671
+ /**
2672
+ * Extension digits to dial after the main number connects. Sent via DTMF. Allow
2673
+ * digits, '\*', '#'.
2674
+ */
2675
+ extension?: string;
2528
2676
  }
2529
2677
  interface TransferDestinationInferred {
2530
2678
  /**
@@ -2930,15 +3078,36 @@ export declare namespace LlmUpdateParams {
2930
3078
  name: string;
2931
3079
  type: 'mcp';
2932
3080
  /**
2933
- * The input schema of the MCP tool.
3081
+ * The description for the sentence agent say during execution. Only applicable
3082
+ * when speak_during_execution is true. Can write what to say or even provide
3083
+ * examples. The default is "The message you will say to callee when calling this
3084
+ * tool. Make sure it fits into the conversation smoothly.".
2934
3085
  */
2935
- input_schema?: {
2936
- [key: string]: string;
2937
- };
3086
+ execution_message_description?: string;
2938
3087
  /**
2939
3088
  * Unique id of the MCP.
2940
3089
  */
2941
3090
  mcp_id?: string;
3091
+ /**
3092
+ * Response variables to add to dynamic variables, key is the variable name, value
3093
+ * is the path to the variable in the response
3094
+ */
3095
+ response_variables?: {
3096
+ [key: string]: string;
3097
+ };
3098
+ /**
3099
+ * Determines whether the agent would call LLM another time and speak when the
3100
+ * result of function is obtained. Usually this needs to get turned on so user can
3101
+ * get update for the function call.
3102
+ */
3103
+ speak_after_execution?: boolean;
3104
+ /**
3105
+ * Determines whether the agent would say sentence like "One moment, let me check
3106
+ * that." when executing the function. Recommend to turn on if your function call
3107
+ * takes over 1s (including network) to complete, so that your agent remains
3108
+ * responsive.
3109
+ */
3110
+ speak_during_execution?: boolean;
2942
3111
  }
2943
3112
  interface SendSMSTool {
2944
3113
  /**
@@ -2972,6 +3141,19 @@ export declare namespace LlmUpdateParams {
2972
3141
  type?: 'inferred';
2973
3142
  }
2974
3143
  }
3144
+ /**
3145
+ * Knowledge base configuration for RAG retrieval.
3146
+ */
3147
+ interface KBConfig {
3148
+ /**
3149
+ * Similarity threshold for filtering search results
3150
+ */
3151
+ filter_score?: number;
3152
+ /**
3153
+ * Max number of knowledge base chunks to retrieve
3154
+ */
3155
+ top_k?: number;
3156
+ }
2975
3157
  interface State {
2976
3158
  /**
2977
3159
  * Name of the state, must be unique for each state. Must be consisted of a-z, A-Z,
@@ -3099,6 +3281,11 @@ export declare namespace LlmUpdateParams {
3099
3281
  * The type of transfer destination.
3100
3282
  */
3101
3283
  type: 'predefined';
3284
+ /**
3285
+ * Extension digits to dial after the main number connects. Sent via DTMF. Allow
3286
+ * digits, '\*', '#'.
3287
+ */
3288
+ extension?: string;
3102
3289
  }
3103
3290
  interface TransferDestinationInferred {
3104
3291
  /**
@@ -3504,15 +3691,36 @@ export declare namespace LlmUpdateParams {
3504
3691
  name: string;
3505
3692
  type: 'mcp';
3506
3693
  /**
3507
- * The input schema of the MCP tool.
3694
+ * The description for the sentence agent say during execution. Only applicable
3695
+ * when speak_during_execution is true. Can write what to say or even provide
3696
+ * examples. The default is "The message you will say to callee when calling this
3697
+ * tool. Make sure it fits into the conversation smoothly.".
3508
3698
  */
3509
- input_schema?: {
3510
- [key: string]: string;
3511
- };
3699
+ execution_message_description?: string;
3512
3700
  /**
3513
3701
  * Unique id of the MCP.
3514
3702
  */
3515
3703
  mcp_id?: string;
3704
+ /**
3705
+ * Response variables to add to dynamic variables, key is the variable name, value
3706
+ * is the path to the variable in the response
3707
+ */
3708
+ response_variables?: {
3709
+ [key: string]: string;
3710
+ };
3711
+ /**
3712
+ * Determines whether the agent would call LLM another time and speak when the
3713
+ * result of function is obtained. Usually this needs to get turned on so user can
3714
+ * get update for the function call.
3715
+ */
3716
+ speak_after_execution?: boolean;
3717
+ /**
3718
+ * Determines whether the agent would say sentence like "One moment, let me check
3719
+ * that." when executing the function. Recommend to turn on if your function call
3720
+ * takes over 1s (including network) to complete, so that your agent remains
3721
+ * responsive.
3722
+ */
3723
+ speak_during_execution?: boolean;
3516
3724
  }
3517
3725
  interface SendSMSTool {
3518
3726
  /**