Sends a prompt for the agent to process and respond to. Use return control event type for function calling.
- *The CLI doesn't support InvokeAgent.
Sends a prompt for the agent to process and respond to. Note the following fields for the request:
*To continue the same conversation with an agent, use the same sessionId value in the request.
End a conversation by setting endSession to true.
In the sessionState object, you can include attributes for the session or prompt or parameters returned from the action group.
Use return control event type for function calling.
+ *In the sessionState object, you can include attributes for the session or prompt or, if you configured an action group to return control, results from invocation of the action group.
The response is returned in the bytes field of the chunk object.
If you set enableTrace to true in the request, you can trace the agent's steps and reasoning process that led it to the response.
If the action predicted was configured to return control, the response returns parameters for the action, elicited from the user, in the returnControl field.
Errors are also surfaced in the response.
*This data type is used in the following API operations:
*In the returnControlInvocationResults field of the Retrieve request
+ *
In the returnControlInvocationResults field of the InvokeAgent request
*
This data type is used in the following API operations:
*In the returnControlInvocationResults of the Retrieve request
+ *
In the returnControlInvocationResults of the InvokeAgent request
*
The response body from the API operation. The key of the object is the content type. The response may be returned directly or from the Lambda function.
+ *The response body from the API operation. The key of the object is the content type (currently, only TEXT is supported). The response may be returned directly or from the Lambda function.
This data type is used in the following API operations:
*In the returnControlInvocationResults of the Retrieve request
+ *
In the returnControlInvocationResults of the InvokeAgent request
*
The response from the function call using the parameters. The response may be returned directly or from the Lambda function.
+ *The response from the function call using the parameters. The key of the object is the content type (currently, only TEXT is supported). The response may be returned directly or from the Lambda function.
A result from the action group invocation.
+ *A result from the invocation of an action. For more information, see Return control to the agent developer and Control session context.
*This data type is used in the following API operations:
*- * Retrieve request + * InvokeAgent request *
*Contains information about the results from the action group invocation.
+ *Contains information about the results from the action group invocation. For more information, see Return control to the agent developer and Control session context.
+ *If you include this field, the inputText field will be ignored.
The identifier of the invocation.
+ *The identifier of the invocation of an action. This value must match the invocationId returned in the InvokeAgent response for the action whose results are provided in the returnControlInvocationResults field. For more information, see Return control to the agent developer and Control session context.
Contains parameters that specify various attributes of the session. For more information, see Control session context.
+ *If you include returnControlInvocationResults in the sessionState field, the inputText field will be ignored.
The prompt text to send the agent.
+ *If you include returnControlInvocationResults in the sessionState field, the inputText field will be ignored.
- * Retrieve response – in the span field
span field
*
- * Retrieve response – in the textResponsePart field
textResponsePart field
*
- * Retrieve response – in the generatedResponsePart field
generatedResponsePart field
* @@ -563,7 +572,7 @@ export interface GeneratedResponsePart { *
- * Retrieve response – in the content field
content field
*
- * Retrieve response – in the s3Location field
s3Location field
*
- * Retrieve response – in the locatino field
locatino field
*
- * Retrieve response – in the retrievedReferences field
retrievedReferences field
*
- * Retrieve response – in the citations field
citations field
* @@ -770,7 +779,7 @@ export class ResourceNotFoundException extends __BaseException { *
- * Retrieve response + * InvokeAgent response *
*- * Retrieve response + * InvokeAgent response *
*This data type is used in the following API operations:
*In the returnControl field of the Retrieve response
+ *
In the returnControl field of the InvokeAgent response
*
This data type is used in the following API operations:
*In the returnControl field of the Retrieve response
+ *
In the returnControl field of the InvokeAgent response
*
This data type is used in the following API operations:
*In the returnControl field of the Retrieve response
+ *
In the returnControl field of the InvokeAgent response
*
Contains details about the API operation or function that the agent predicts should be called.
+ *Contains details about the API operation or function that the agent predicts should be called.
*This data type is used in the following API operations:
*In the returnControl field of the Retrieve response
+ *
In the returnControl field of the InvokeAgent response
*
- * Retrieve response + * InvokeAgent response *
*The configuration details for the guardrail.
+ * @public + */ +export interface GuardrailConfiguration { + /** + *The unique identifier for the guardrail.
+ * @public + */ + guardrailId: string | undefined; + + /** + *The version of the guardrail.
+ * @public + */ + guardrailVersion: string | undefined; +} + +/** + *Configuration settings for text generation using a language model via the + * RetrieveAndGenerate operation. Includes parameters like temperature, top-p, maximum token + * count, and stop sequences.
+ *The valid range of maxTokens depends on the accepted values for your chosen
+ * model's inference parameters. To see the inference parameters for your model, see Inference
+ * parameters for foundation models.
+ *
Controls the random-ness of text generated by the language model, influencing how much the model sticks to the most predictable next words versus exploring more surprising options. A lower temperature value (e.g. 0.2 or 0.3) makes model outputs more deterministic or predictable, while a higher temperature (e.g. 0.8 or 0.9) makes the outputs more creative or unpredictable.
+ * @public + */ + temperature?: number; + + /** + *A probability distribution threshold which controls what the model considers for the set of possible next tokens. The model will only consider the top p% of the probability distribution when generating the next token.
+ * @public + */ + topP?: number; + + /** + *The maximum number of tokens to generate in the output text. Do not use the minimum of 0 + * or the maximum of 65536. The limit values described here are arbitary values, for actual + * values consult the limits defined by your specific model.
+ * @public + */ + maxTokens?: number; + + /** + *A list of sequences of characters that, if generated, will cause the model to stop + * generating further tokens. Do not use a minimum length of 1 or a maximum length of 1000. The + * limit values described here are arbitary values, for actual values consult the limits defined + * by your specific model.
+ * @public + */ + stopSequences?: string[]; +} + +/** + *The configuration for inference settings when generating responses using RetrieveAndGenerate.
+ * @public + */ +export interface InferenceConfig { + /** + *Configuration settings specific to text generation while generating responses using RetrieveAndGenerate.
+ * @public + */ + textInferenceConfig?: TextInferenceConfig; +} + /** *Contains the template for the prompt that's sent to the model for response generation. For more information, see Knowledge base prompt templates.
*This data type is used in the following API operations:
@@ -2274,6 +2356,24 @@ export interface ExternalSourcesGenerationConfiguration { * @public */ promptTemplate?: PromptTemplate; + + /** + *The configuration details for the guardrail.
+ * @public + */ + guardrailConfiguration?: GuardrailConfiguration; + + /** + *Configuration settings for inference when using RetrieveAndGenerate to generate responses while using an external source.
+ * @public + */ + inferenceConfig?: InferenceConfig; + + /** + *Additional model parameters and their corresponding values not included in the textInferenceConfig structure for an external source. Takes in custom model parameters specific to the language model being used.
+ * @public + */ + additionalModelRequestFields?: RecordThe configuration details for the guardrail.
+ * @public + */ + guardrailConfiguration?: GuardrailConfiguration; + + /** + *Configuration settings for inference when using RetrieveAndGenerate to generate responses while using a knowledge base as a source.
+ * @public + */ + inferenceConfig?: InferenceConfig; + + /** + *Additional model parameters and corresponding values not included in the textInferenceConfig structure for a knowledge base. This allows users to provide custom model parameters specific to the language model being used.
+ * @public + */ + additionalModelRequestFields?: RecordContains the response generated from querying the knowledge base.
*This data type is used in the following API operations:
@@ -2507,6 +2639,12 @@ export interface RetrieveAndGenerateResponse { * @public */ citations?: Citation[]; + + /** + *Specifies if there is a guardrail intervention in the response.
+ * @public + */ + guardrailAction?: GuadrailAction; } /** diff --git a/clients/client-bedrock-agent-runtime/src/protocols/Aws_restJson1.ts b/clients/client-bedrock-agent-runtime/src/protocols/Aws_restJson1.ts index fcd68151ecbc..d29305fee964 100644 --- a/clients/client-bedrock-agent-runtime/src/protocols/Aws_restJson1.ts +++ b/clients/client-bedrock-agent-runtime/src/protocols/Aws_restJson1.ts @@ -19,6 +19,7 @@ import { limitedParseFloat32 as __limitedParseFloat32, map, resolvedPath as __resolvedPath, + serializeFloat as __serializeFloat, take, withBaseException, } from "@smithy/smithy-client"; @@ -53,6 +54,8 @@ import { FilterAttribute, FunctionResult, GenerationConfiguration, + GuardrailConfiguration, + InferenceConfig, InferenceConfiguration, InternalServerException, InvocationResultMember, @@ -80,6 +83,7 @@ import { S3ObjectDoc, ServiceQuotaExceededException, SessionState, + TextInferenceConfig, ThrottlingException, Trace, TracePart, @@ -222,6 +226,7 @@ export const de_RetrieveAndGenerateCommand = async ( const data: RecordContains information about the API operation that the agent predicts should be called.
\nThis data type is used in the following API operations:
\nIn the returnControl field of the Retrieve response\n
Contains information about the API operation that the agent predicts should be called.
\nThis data type is used in the following API operations:
\nIn the returnControl field of the InvokeAgent response\n
Information about a parameter to provide to the API request.
\nThis data type is used in the following API operations:
\n\n Retrieve response\n
\nInformation about a parameter to provide to the API request.
\nThis data type is used in the following API operations:
\n\n InvokeAgent response\n
\nThe request body to provide for the API request, as the agent elicited from the user.
\nThis data type is used in the following API operations:
\n\n Retrieve response\n
\nThe request body to provide for the API request, as the agent elicited from the user.
\nThis data type is used in the following API operations:
\n\n InvokeAgent response\n
\nThe response body from the API operation. The key of the object is the content type. The response may be returned directly or from the Lambda function.
" + "smithy.api#documentation": "The response body from the API operation. The key of the object is the content type (currently, only TEXT is supported). The response may be returned directly or from the Lambda function.
Contains information about the API operation that was called from the action group and the response body that was returned.
\nThis data type is used in the following API operations:
\nIn the returnControlInvocationResults of the Retrieve request\n
Contains information about the API operation that was called from the action group and the response body that was returned.
\nThis data type is used in the following API operations:
\nIn the returnControlInvocationResults of the InvokeAgent request\n
An object containing a segment of the generated response that is based on a source in the knowledge base, alongside information about the source.
\nThis data type is used in the following API operations:
\n\n Retrieve response – in the citations field
\n RetrieveAndGenerate response – in the citations field
An object containing a segment of the generated response that is based on a source in the knowledge base, alongside information about the source.
\nThis data type is used in the following API operations:
\n\n InvokeAgent response – in the citations field
\n RetrieveAndGenerate response – in the citations field
Contains the body of the API response.
\nThis data type is used in the following API operations:
\nIn the returnControlInvocationResults field of the Retrieve request\n
Contains the body of the API response.
\nThis data type is used in the following API operations:
\nIn the returnControlInvocationResults field of the InvokeAgent request\n
Contain the textPromptTemplate string for the external source wrapper object.
" } + }, + "guardrailConfiguration": { + "target": "com.amazonaws.bedrockagentruntime#GuardrailConfiguration", + "traits": { + "smithy.api#documentation": "The configuration details for the guardrail.
" + } + }, + "inferenceConfig": { + "target": "com.amazonaws.bedrockagentruntime#InferenceConfig", + "traits": { + "smithy.api#documentation": "Configuration settings for inference when using RetrieveAndGenerate to generate responses while using an external source.
" + } + }, + "additionalModelRequestFields": { + "target": "com.amazonaws.bedrockagentruntime#AdditionalModelRequestFields", + "traits": { + "smithy.api#documentation": "Additional model parameters and their corresponding values not included in the textInferenceConfig structure for an external source. Takes in custom model parameters specific to the language model being used.
" + } } }, "traits": { @@ -1345,7 +1384,7 @@ } }, "traits": { - "smithy.api#documentation": "Contains information about the function that the agent predicts should be called.
\nThis data type is used in the following API operations:
\nIn the returnControl field of the Retrieve response\n
Contains information about the function that the agent predicts should be called.
\nThis data type is used in the following API operations:
\nIn the returnControl field of the InvokeAgent response\n
Contains information about a parameter of the function.
\nThis data type is used in the following API operations:
\nIn the returnControl field of the Retrieve response\n
Contains information about a parameter of the function.
\nThis data type is used in the following API operations:
\nIn the returnControl field of the InvokeAgent response\n
The response from the function call using the parameters. The response may be returned directly or from the Lambda function.
" + "smithy.api#documentation": "The response from the function call using the parameters. The key of the object is the content type (currently, only TEXT is supported). The response may be returned directly or from the Lambda function.
Contains information about the function that was called from the action group and the response that was returned.
\nThis data type is used in the following API operations:
\nIn the returnControlInvocationResults of the Retrieve request\n
Contains information about the function that was called from the action group and the response that was returned.
\nThis data type is used in the following API operations:
\nIn the returnControlInvocationResults of the InvokeAgent request\n
Contains metadata about a part of the generated response that is accompanied by a citation.
\nThis data type is used in the following API operations:
\n\n Retrieve response – in the generatedResponsePart field
\n RetrieveAndGenerate response – in the generatedResponsePart field
Contains metadata about a part of the generated response that is accompanied by a citation.
\nThis data type is used in the following API operations:
\n\n InvokeAgent response – in the generatedResponsePart field
\n RetrieveAndGenerate response – in the generatedResponsePart field
Contains the template for the prompt that's sent to the model for response generation.
" } + }, + "guardrailConfiguration": { + "target": "com.amazonaws.bedrockagentruntime#GuardrailConfiguration", + "traits": { + "smithy.api#documentation": "The configuration details for the guardrail.
" + } + }, + "inferenceConfig": { + "target": "com.amazonaws.bedrockagentruntime#InferenceConfig", + "traits": { + "smithy.api#documentation": "Configuration settings for inference when using RetrieveAndGenerate to generate responses while using a knowledge base as a source.
" + } + }, + "additionalModelRequestFields": { + "target": "com.amazonaws.bedrockagentruntime#AdditionalModelRequestFields", + "traits": { + "smithy.api#documentation": "Additional model parameters and corresponding values not included in the textInferenceConfig structure for a knowledge base. This allows users to provide custom model parameters specific to the language model being used.
" + } } }, "traits": { "smithy.api#documentation": "Contains configurations for response generation based on the knowledge base query results.
\nThis data type is used in the following API operations:
\nThe unique identifier for the guardrail.
", + "smithy.api#length": { + "max": 64 + }, + "smithy.api#pattern": "^[a-z0-9]+$", + "smithy.api#required": {} + } + }, + "guardrailVersion": { + "target": "smithy.api#String", + "traits": { + "smithy.api#documentation": "The version of the guardrail.
", + "smithy.api#length": { + "min": 1, + "max": 5 + }, + "smithy.api#pattern": "^(([1-9][0-9]{0,7})|(DRAFT))$", + "smithy.api#required": {} + } + } + }, + "traits": { + "smithy.api#documentation": "The configuration details for the guardrail.
" + } + }, "com.amazonaws.bedrockagentruntime#Identifier": { "type": "string", "traits": { @@ -1451,6 +1556,20 @@ "smithy.api#sensitive": {} } }, + "com.amazonaws.bedrockagentruntime#InferenceConfig": { + "type": "structure", + "members": { + "textInferenceConfig": { + "target": "com.amazonaws.bedrockagentruntime#TextInferenceConfig", + "traits": { + "smithy.api#documentation": "Configuration settings specific to text generation while generating responses using RetrieveAndGenerate.
" + } + } + }, + "traits": { + "smithy.api#documentation": "The configuration for inference settings when generating responses using RetrieveAndGenerate.
" + } + }, "com.amazonaws.bedrockagentruntime#InferenceConfiguration": { "type": "structure", "members": { @@ -1569,7 +1688,7 @@ } }, "traits": { - "smithy.api#documentation": "Contains details about the API operation or function that the agent predicts should be called.
\nThis data type is used in the following API operations:
\nIn the returnControl field of the Retrieve response\n
Contains details about the API operation or function that the agent predicts should be called.
\nThis data type is used in the following API operations:
\nIn the returnControl field of the InvokeAgent response\n
A result from the action group invocation.
\nThis data type is used in the following API operations:
\n\n Retrieve request\n
\nA result from the invocation of an action. For more information, see Return control to the agent developer and Control session context.
\nThis data type is used in the following API operations:
\n\n InvokeAgent request\n
\nSends a prompt for the agent to process and respond to. Use return control event type for function calling.
\nThe CLI doesn't support InvokeAgent.
To continue the same conversation with an agent, use the same sessionId value in the request.
To activate trace enablement, turn enableTrace to true. Trace enablement helps you follow the agent's reasoning process that led it to the information it processed, the actions it took, and the final result it yielded. For more information, see Trace enablement.
End a conversation by setting endSession to true.
In the sessionState object, you can include attributes for the session or prompt or parameters returned from the action group.
Use return control event type for function calling.
\nThe response is returned in the bytes field of the chunk object.
The attribution object contains citations for parts of the response.
If you set enableTrace to true in the request, you can trace the agent's steps and reasoning process that led it to the response.
Errors are also surfaced in the response.
\nThe CLI doesn't support InvokeAgent.
Sends a prompt for the agent to process and respond to. Note the following fields for the request:
\nTo continue the same conversation with an agent, use the same sessionId value in the request.
To activate trace enablement, turn enableTrace to true. Trace enablement helps you follow the agent's reasoning process that led it to the information it processed, the actions it took, and the final result it yielded. For more information, see Trace enablement.
End a conversation by setting endSession to true.
In the sessionState object, you can include attributes for the session or prompt or, if you configured an action group to return control, results from invocation of the action group.
The response is returned in the bytes field of the chunk object.
The attribution object contains citations for parts of the response.
If you set enableTrace to true in the request, you can trace the agent's steps and reasoning process that led it to the response.
If the action predicted was configured to return control, the response returns parameters for the action, elicited from the user, in the returnControl field.
Errors are also surfaced in the response.
\nContains parameters that specify various attributes of the session. For more information, see Control session context.
" + "smithy.api#documentation": "Contains parameters that specify various attributes of the session. For more information, see Control session context.
\nIf you include returnControlInvocationResults in the sessionState field, the inputText field will be ignored.
The prompt text to send the agent.
" + "smithy.api#documentation": "The prompt text to send the agent.
\nIf you include returnControlInvocationResults in the sessionState field, the inputText field will be ignored.
Contains the parameters in the request body.
" } }, + "com.amazonaws.bedrockagentruntime#RAGStopSequences": { + "type": "list", + "member": { + "target": "smithy.api#String", + "traits": { + "smithy.api#length": { + "min": 1, + "max": 1000 + } + } + }, + "traits": { + "smithy.api#length": { + "min": 0, + "max": 4 + } + } + }, "com.amazonaws.bedrockagentruntime#Rationale": { "type": "structure", "members": { @@ -2669,7 +2815,7 @@ } }, "traits": { - "smithy.api#documentation": "Contains the cited text from the data source.
\nThis data type is used in the following API operations:
\n\n Retrieve response – in the content field
\n RetrieveAndGenerate response – in the content field
\n Retrieve response – in the content field
Contains the cited text from the data source.
\nThis data type is used in the following API operations:
\n\n Retrieve response – in the content field
\n RetrieveAndGenerate response – in the content field
\n InvokeAgent response – in the content field
Contains information about the location of the data source.
\nThis data type is used in the following API operations:
\n\n Retrieve response – in the location field
\n RetrieveAndGenerate response – in the location field
\n Retrieve response – in the locatino field
Contains information about the location of the data source.
\nThis data type is used in the following API operations:
\n\n Retrieve response – in the location field
\n RetrieveAndGenerate response – in the location field
\n InvokeAgent response – in the locatino field
Contains the S3 location of the data source.
\nThis data type is used in the following API operations:
\n\n Retrieve response – in the s3Location field
\n RetrieveAndGenerate response – in the s3Location field
\n Retrieve response – in the s3Location field
Contains the S3 location of the data source.
\nThis data type is used in the following API operations:
\n\n Retrieve response – in the s3Location field
\n RetrieveAndGenerate response – in the s3Location field
\n InvokeAgent response – in the s3Location field
A list of segments of the generated response that are based on sources in the knowledge base, alongside information about the sources.
" } + }, + "guardrailAction": { + "target": "com.amazonaws.bedrockagentruntime#GuadrailAction", + "traits": { + "smithy.api#documentation": "Specifies if there is a guardrail intervention in the response.
" + } } }, "traits": { @@ -3095,7 +3247,7 @@ } }, "traits": { - "smithy.api#documentation": "Contains metadata about a source cited for the generated response.
\nThis data type is used in the following API operations:
\n\n RetrieveAndGenerate response – in the retrievedReferences field
\n Retrieve response – in the retrievedReferences field
Contains metadata about a source cited for the generated response.
\nThis data type is used in the following API operations:
\n\n RetrieveAndGenerate response – in the retrievedReferences field
\n InvokeAgent response – in the retrievedReferences field
Contains information to return from the action group that the agent has predicted to invoke.
\nThis data type is used in the following API operations:
\n\n Retrieve response\n
\nContains information to return from the action group that the agent has predicted to invoke.
\nThis data type is used in the following API operations:
\n\n InvokeAgent response\n
\nContains information about the results from the action group invocation.
" + "smithy.api#documentation": "Contains information about the results from the action group invocation. For more information, see Return control to the agent developer and Control session context.
\nIf you include this field, the inputText field will be ignored.
The identifier of the invocation.
" + "smithy.api#documentation": "The identifier of the invocation of an action. This value must match the invocationId returned in the InvokeAgent response for the action whose results are provided in the returnControlInvocationResults field. For more information, see Return control to the agent developer and Control session context.
Contains information about where the text with a citation begins and ends in the generated output.
\nThis data type is used in the following API operations:
\n\n RetrieveAndGenerate response – in the span field
\n Retrieve response – in the span field
Contains information about where the text with a citation begins and ends in the generated output.
\nThis data type is used in the following API operations:
\n\n RetrieveAndGenerate response – in the span field
\n InvokeAgent response – in the span field
Controls the random-ness of text generated by the language model, influencing how much the model sticks to the most predictable next words versus exploring more surprising options. A lower temperature value (e.g. 0.2 or 0.3) makes model outputs more deterministic or predictable, while a higher temperature (e.g. 0.8 or 0.9) makes the outputs more creative or unpredictable.
" + } + }, + "topP": { + "target": "com.amazonaws.bedrockagentruntime#TopP", + "traits": { + "smithy.api#documentation": "A probability distribution threshold which controls what the model considers for the set of possible next tokens. The model will only consider the top p% of the probability distribution when generating the next token.
" + } + }, + "maxTokens": { + "target": "com.amazonaws.bedrockagentruntime#MaxTokens", + "traits": { + "smithy.api#documentation": "The maximum number of tokens to generate in the output text. Do not use the minimum of 0\n or the maximum of 65536. The limit values described here are arbitary values, for actual\n values consult the limits defined by your specific model.
" + } + }, + "stopSequences": { + "target": "com.amazonaws.bedrockagentruntime#RAGStopSequences", + "traits": { + "smithy.api#documentation": "A list of sequences of characters that, if generated, will cause the model to stop\n generating further tokens. Do not use a minimum length of 1 or a maximum length of 1000. The\n limit values described here are arbitary values, for actual values consult the limits defined\n by your specific model.
" + } + } + }, + "traits": { + "smithy.api#documentation": "Configuration settings for text generation using a language model via the\n RetrieveAndGenerate operation. Includes parameters like temperature, top-p, maximum token\n count, and stop sequences.
\nThe valid range of maxTokens depends on the accepted values for your chosen\n model's inference parameters. To see the inference parameters for your model, see Inference\n parameters for foundation models.\n
Contains the part of the generated text that contains a citation, alongside where it begins and ends.
\nThis data type is used in the following API operations:
\n\n RetrieveAndGenerate response – in the textResponsePart field
\n Retrieve response – in the textResponsePart field
Contains the part of the generated text that contains a citation, alongside where it begins and ends.
\nThis data type is used in the following API operations:
\n\n RetrieveAndGenerate response – in the textResponsePart field
\n InvokeAgent response – in the textResponsePart field