Generative Ai Agent Runtime¶
|
OCI Generative AI Agents is a fully managed service that combines the power of large language models (LLMs) with an intelligent retrieval system to create contextually relevant answers by searching your knowledge base, making your AI applications smart and efficient. |
|
This class provides a wrapper around |
Models¶
Chat details for managing user interactions and tool executions. |
|
The response of a chat request. |
|
The source of information for the agent's response. |
|
Information about the new session. |
|
The trace information about the error. |
|
Contains trace information related to execution of tool. |
|
Represent a function call the client need to execute in order to fulfill the required action. |
|
|
Represents an action for recording the result of a function call. |
|
Represents an action for submitting function call outputs. |
The trace information about the generated response. |
|
Represents a generic tool input schema that accepts flexible, freeform JSON parameters. |
|
Specifies an output format for tools that return freeform or unstructured JSON content. |
|
|
An action describing user approval to the agent. |
|
An object describing human confirmation of tool execution that is required from the user. |
|
Represents metadata about a field including its name, type, supported operations, and possible values. |
|
Represents a metadata value and the number of appearances in knowledge base. |
The Message resource generated by the agent. |
|
The content of the message. |
|
Filter for retrieving specific metadata |
|
Model details. |
|
Details of model parameters. |
|
|
The location of the OCI Database that the agent will use. |
|
The OCI Object Storage that contains the data files. |
|
The location of the OCI Search with OpenSearch that the agent will use. |
The paragraph of the generated message that contains a citation. |
|
The source of information for the paragraph of agent's response. |
|
The base structure for an action that has already been performed. |
|
Contains trace information related to planning at the platform. |
|
Specifies the output format for RAG tool, including the generated answer and citations. |
|
Represents an action that needs to be performed by the user or client. |
|
The trace that displays the retrieval information. |
|
|
Details to retrieve metadata |
A session represents an interactive conversation initiated by a user through an API to engage with an agent. |
|
The details of source of the event. |
|
The location of the data files that the agent will use. |
|
Specifies file locations for SQL result data. |
|
Specifies the output format for SQL tools, including the generated SQL statement and an optional preview of the query result. |
|
Base object for tool input parameters. |
|
The trace information about the tool selection from multiple tools. |
|
Base schema for tool outputs. |
|
The trace that displays the internal progression, such as reasoning and actions during an execution. |
|
Information about the updated session. |
|
Details of usage. |
|
Usage details. |