DBMS_CLOUD_AI_AGENT Package

This package defines and manages Select AI agents, tasks, tools, and orchestration.

• Summary of DBMS_CLOUD_AI_AGENT Subprograms
  This section covers the Select AI Agent's DBMS_CLOUD_AI_AGENT subprograms provided with Autonomous AI Database.

See Also:

• Select AI Agent

• DBMS_CLOUD_AI_AGENT Views

Summary of DBMS_CLOUD_AI_AGENT Subprograms

This section covers the Select AI Agent's DBMS_CLOUD_AI_AGENT subprograms provided with Autonomous AI Database.

Subprogram	Description
CREATE_AGENT Procedure	This procedure creates an agent.
CREATE_AGENT Attributes	Provides attributes for creating an agent.
ENABLE_AGENT Procedure	This procedure enables an agent.
DISABLE_AGENT Procedure	This procedure disables an agent.
DROP_AGENT Procedure	This procedure drops an existing agent.
CREATE_TASK Procedure	This procedure creates a task that an agent and agent team can include.
CREATE_TASK Attributes	Provides attributes for creating an agent task.
ENABLE_TASK Procedure	This procedure enables an agent task.
DISABLE_TASK Procedure	This procedure disables an agent task.
DROP_TASK Procedure	This procedure drops an existing agent task.
CREATE_TOOL Procedure	This procedure creates custom tools that an agent can include.
CREATE_TOOL Attributes	Provides attributes for creating tools.
ENABLE_TOOL Procedure	This procedure enables a registered tool.
DISABLE_TOOL Procedure	This procedure disables a registered tool.
DROP_TOOL Procedure	This procedure drops an existing tool.
CREATE_TEAM Procedure	This procedure creates an agent team and includes agents and tasks.
CREATE_TEAM Attributes	Provides attributes for creating an agent team.
SET_TEAM Function	This procedure sets the agent team in the current session.
GET_TEAM Function	This procedure returns the agent team set in the current session.
CLEAR_TEAM Function	This procedure clears the agent team set in the current session.
RUN_TEAM Function	This procedure creates a new agent or runs a paused one.
ENABLE_TEAM Procedure	This procedure enables an agent team.
DISABLE_TEAM Procedure	This procedure disables an agent team.
DROP_TEAM Procedure	This procedure drops an existing agent team.

• CREATE_AGENT Procedure
  Use the DBMS_CLOUD_AI_AGENT.CREATE_AGENT procedure to register a new AI agent in the Select AI Agent framework. Define the agent’s identity using agent_name, configure its behavior with the attributes parameter, and optionally provide a description.
• CREATE_AGENT Attributes
  These attributes define an agent's role and the AI profile name.
• ENABLE_AGENT Procedure
  This procedure enables the AI agent that the user specifies. The procedure changes the status of the AI agent to ENABLED.
• DISABLE_AGENT Procedure
  This procedure disables the AI agent in the current database. The status of the AI agent is changed to DISABLED by this procedure.
• DROP_AGENT Procedure
  The procedure drops an existing AI agent. If the agent does not exist, then the procedure throws an error.
• CREATE_TASK Procedure
  Use the DBMS_CLOUD_AI_AGENT.CREATE_TASK procedure to define a task that a Select AI agent can include in its reasoning process. Each task has a unique name and a set of attributes that specify the agent’s behavior when planning and performing the task.
• CREATE_TASK Attributes
  These attributes define agent's instructions and tools available for the task.
• ENABLE_TASK Procedure
  Use the DBMS_CLOUD_AI_AGENT.ENABLE_TASK procedure to enable a previously defined task. This changes the task’s status to ENABLED, allowing the agent to use it during processing.
• DISABLE_TASK Procedure
  Use the DBMS_CLOUD_AI_AGENT.DISABLE_TASK procedure to disable a task in the current database. This changes the task’s status to DISABLED, preventing the agent from using it during processing.
• DROP_TASK Procedure
  Use the DBMS_CLOUD_AI_AGENT.DROP_TASK procedure to remove a previously defined task from the database. If the task does not exist, you can suppress the error by setting the force parameter to TRUE.
• CREATE_TOOL Procedure
  Use the DBMS_CLOUD_AI_AGENT.CREATE_TOOL procedure to register a custom tool that an agent can use during task processing. Each tool is identified by a unique tool_name and includes attributes that define its purpose, implementation logic, and metadata.
• CREATE_TOOL Attributes
  These attributes provide tool configuration using a JSON object in the attributes parameter. Each attribute configures how the agent uses the tool during reasoning.
• ENABLE_TOOL Procedure
  Use the DBMS_CLOUD_AI_AGENT.ENABLE_TOOL procedure to enable a previously registered tool. This changes the tool's status to ENABLED, allowing the agent to use it tasks.
• DISABLE_TOOL Procedure
  Use the DBMS_CLOUD_AI_AGENT.DISABLE_TOOL procedure to disable a registered tool. This changes the tool's status to DISABLED, preventing the agent from using it in tasks.
• DROP_TOOL Procedure
  Use the DBMS_CLOUD_AI_AGENT.DROP_TOOL procedure to remove a previously registered tool with the agent. If the task does not exist, you can suppress the error by setting the force parameter to TRUE.
• CREATE_TEAM Procedure
  Use the DBMS_CLOUD_AI_AGENT.CREATE_TEAM procedure to define a team of AI agents that work together to accomplish tasks. You specify agents and their associated tasks, along with how the tasks should be processed.
• CREATE_TEAM Attributes
  These attributes provide AI agent team configuration. Each team has a unique team_name and uses a set of attributes to define its structure and execution strategy.
• SET_TEAM Function
  Use DBMS_CLOUD_AI_AGENT.SET_TEAM to specify the AI agent team for the current session.
• GET_TEAM Function
  Use DBMS_CLOUD_AI_AGENT.GET_TEAM to return the AI agent team for the current session.
• CLEAR_TEAM Function
  Use DBMS_CLOUD_AI_AGENT.CLEAR_TEAM to clear the AI agent team set in the current session. This procedure does not drop the team.
• RUN_TEAM Function
  Use DBMS_CLOUD_AI_AGENT.RUN_TEAM to start a new AI agent team or resume a paused one that is waiting for human input. If you provide an existing process ID and the associated team process is in the WAITING_FOR_HUMAN state, the function resumes the workflow using the input you provide as the human response.
• ENABLE_TEAM Procedure
  Use the DBMS_CLOUD_AI_AGENT.ENABLE_TEAM procedure to enable a user-specified AI agent team. This changes the team's status to ENABLED.
• DISABLE_TEAM Procedure
  Use the DBMS_CLOUD_AI_AGENT.DISABLE_TEAM procedure to disable an existing AI agent team. This changes the team's status to DISABLED.
• DROP_TEAM Procedure
  Use the DBMS_CLOUD_AI_AGENT.DROP_TEAM procedure to remove a previously created AI agent team.

CREATE_AGENT Procedure

Use the DBMS_CLOUD_AI_AGENT.CREATE_AGENT procedure to register a new AI agent in the Select AI Agent framework. Define the agent’s identity using agent_name, configure its behavior with the attributes parameter, and optionally provide a description.

Syntax

DBMS_CLOUD_AI_AGENT.CREATE_AGENT(
     agent_name          IN  VARCHAR2,
     attributes          IN  CLOB,
     status              IN  VARCHAR2  DEFAULT NULL,
     description         IN  CLOB      DEFAULT NULL
  );

Parameters

Parameter	Description
agent_name	A name for the AI agent. The agent name must follow the naming rules of Oracle SQL identifier. Maximum length of name is 125 characters. This is a mandatory parameter.
attributes	Custom attributes for the AI agent. See CREATE_AGENT Attributes for more details.
status	Status of the agent. The default value is enabled.
description	User-specified description for the AI agent. The default value is NULL.

Example

The following example creates an agent to process customer product returns. For a complete example, see Example: Create a Product Return Agent.

BEGIN
  DBMS_CLOUD_AI_AGENT.CREATE_AGENT(
    agent_name => 'CustomerAgent',
    attributes => '{
       "profile_name": "GOOGLE",
                       "role": "You are an experienced customer agent who deals with customers return request."
     }'
  );
END;
/  
 
PL/SQL procedure successfully completed.

CREATE_AGENT Attributes

These attributes define an agent's role and the AI profile name.

Attributes

Attribute Name	Default Value	Description
profile_name	-	The AI profile that the agent is using to send request to LLM. This is a mandatory parameter.
role	NULL	Define the agent’s function and provide context to the agent. This is sent to LLM. This is a mandatory parameter.
enable_human_tool	True	Enable agent to ask questions to the user for information or clarification.

The following example shows you how you can customize agent attributes in the procedure.

BEGIN
  DBMS_CLOUD_AI_AGENT.CREATE_AGENT(
    agent_name => 'Customer_Return_Agent',
    attributes => '{
       "profile_name": "GOOGLE",
                       "role": "You are an experienced customer agent who deals with customers return request.",
			   "enable_human_tool": "False"
     }'
  );
END;
/  
 
PL/SQL procedure successfully completed

ENABLE_AGENT Procedure

This procedure enables the AI agent that the user specifies. The procedure changes the status of the AI agent to ENABLED.

Syntax

DBMS_CLOUD_AI_AGENT.ENABLE_AGENT(
   agent_name           IN  VARCHAR2
);

Parameters

Parameter	Description
agent_name	Name for the AI agent to enable. This parameter is mandatory.

Example to Enable AI Profile

BEGIN
     DBMS_CLOUD_AI_AGENT.ENABLE_AGENT(
         agent_name    => 'Customer_Return_Agent'
     );
END;
/

DISABLE_AGENT Procedure

This procedure disables the AI agent in the current database. The status of the AI agent is changed to DISABLED by this procedure.

Syntax

DBMS_CLOUD_AI_AGENT.DISABLE_AGENT(
   agent_name           IN  VARCHAR2
);

Parameters

Parameter	Description
agent_name	Name for the AI agent. This parameter is mandatory.

Example

BEGIN
     DBMS_CLOUD_AI_AGENT.DISABLE_AGENT(
         agent_name    => 'Customer_Return_Agent'
     );
END;
/

DROP_AGENT Procedure

The procedure drops an existing AI agent. If the agent does not exist, then the procedure throws an error.

Syntax

DBMS_CLOUD_AI_AGENT.DROP_AGENT(
    agent_name          IN  VARCHAR2,
    force               IN  BOOLEAN DEFAULT FALSE
 );

Parameters

Parameter	Description
agent_name	Name of the AI agent. This is a mandatory parameter.
force	If TRUE, then the procedure ignores errors if AI agent does not exist. The default value for this parameter is FALSE.

Example

BEGIN
     DBMS_CLOUD_AI_AGENT.DROP_AGENT(agent_name => 'Customer_Return_Agent);
END;
/

Usage Notes

Use force to drop an agent and ignore errors if AI agent does not exist.

CREATE_TASK Procedure

Use the DBMS_CLOUD_AI_AGENT.CREATE_TASK procedure to define a task that a Select AI agent can include in its reasoning process. Each task has a unique name and a set of attributes that specify the agent’s behavior when planning and performing the task.

You can define dependencies between tasks using the input attribute, allowing Select AI to pass the output of one task into another. This supports chained reasoning and multi-step workflows.

Syntax

DBMS_CLOUD_AI_AGENT.CREATE_TASK(
     task_name           IN  VARCHAR2,
     attributes          IN  CLOB,
     status              IN  VARCHAR2  DEFAULT NULL,
     description         IN  CLOB      DEFAULT NULL
  );

Parameters

Parameter	Description
task_name	A unique name for the task. This is a mandatory parameter.
attributes	JSON object containing task-specific attributes. See CREATE_TASK Attributes.
status	Status of the task. The default value is ENABLED.
description	User-defined description to help identify the task. This value is stored in the database but not sent to the LLM. The default value is NULL.
enable_human_tool	Enable agent to ask questions to user when it requires information or clarification during a task. Default value is true. Note: Task enable_human_tool overwrites the agent enable_human_tool attribute. If agent enable_human_tool is set to false and task enable_human_tool is set to true, during the task run, the LLM is expected to ask questions.

Example

The following example creates a task to process a product return from a customer.

BEGIN
  DBMS_CLOUD_AI_AGENT.CREATE_TASK(
    task_name => 'Return_And_Price_Match',
    attributes => '{"instruction": "Process a product return request from a customer:{query}' ||
                    '1. Ask customer the reason for return (price match or defective)' ||
                    '2. If price match:' ||
                    '   a. Request customer to provide a price match link' ||
                    '   b. Use websearch tool to get the price for that price match link' ||
                    '   c. Ask customer if they want a refund.' ||
                    '   d. Send email notification only if customer accept the refund.' ||
                    '3. If defective:' ||
                     '   a. Process the defective return",
                    "tools": ["WebSearch", "Email"]}'
  );
END;
/

CREATE_TASK Attributes

These attributes define agent's instructions and tools available for the task.

Attributes

Attribute Name	Default Value	Description
instruction	-	A clear, concise statement that describes what the task should accomplish. This text is included in the prompt sent to the LLM. You can include a {query} placeholder to represent the your natural language input. For example: SELECT AI AGENT <user_prompt> This is a mandatory parameter.
tools	-	A JSON array of tool names that the agent can use to complete the task. For example: "tools": ["RAG", "SQL", "WEBSEARCH", "NOTIFICATION"]. This is a mandatory parameter.
input	-	Identifies other tasks whose outputs should be used as input for this task. Select AI automatically provides these outputs to the LLM.

ENABLE_TASK Procedure

Use the DBMS_CLOUD_AI_AGENT.ENABLE_TASK procedure to enable a previously defined task. This changes the task’s status to ENABLED, allowing the agent to use it during processing.

Syntax

DBMS_CLOUD_AI_AGENT.ENABLE_TASK(
   task_name           IN  VARCHAR2
);

Parameters

Parameter	Description
task_name	Name for the task to enable. This parameter is mandatory.

Usage Notes

The task must already exist in the system. This procedure updates the task status, making it available for agent process during reasoning workflows.

Example to Enable a Task

BEGIN
     DBMS_CLOUD_AI_AGENT.ENABLE_TASK(
         task_name    => 'Return_And_Price_Match'
     );
END;
/

DISABLE_TASK Procedure

Use the DBMS_CLOUD_AI_AGENT.DISABLE_TASK procedure to disable a task in the current database. This changes the task’s status to DISABLED, preventing the agent from using it during processing.

Syntax

DBMS_CLOUD_AI_AGENT.DISABLE_TASK(
   task_name           IN  VARCHAR2
);

Parameters

Parameter	Description
task_name	Name of the task to disable. This parameter is mandatory.

Usage Notes

Disabling a task does not delete it; you can re-enable it later using DBMS_CLOUD_AI_AGENT.ENABLE_TASK. Once disabled, the task becomes inactive and the agent no longer considers it during task selection.

Example

The following example disables an existing task.

BEGIN
     DBMS_CLOUD_AI_AGENT.DISABLE_TASK(
         task_name    => 'Return_And_Price_Match'
     );
END;
/

DROP_TASK Procedure

Use the DBMS_CLOUD_AI_AGENT.DROP_TASK procedure to remove a previously defined task from the database. If the task does not exist, you can suppress the error by setting the force parameter to TRUE.

Syntax

DBMS_CLOUD_AI_AGENT.DROP_TASK(
   task_name IN  VARCHAR2,
   force     IN  BOOLEAN DEFAULT FALSE
);

Parameters

Parameter	Description
task_name	Name of the task to drop. This is a mandatory parameter.
force	If TRUE, then the procedure ignores errors if the task does not exist. The default value for this parameter is FALSE.

Example

BEGIN
     DBMS_CLOUD_AI_AGENT.DROP_TASK(task_name => 'Return_And_Price_Match');
END;
/

Usage Notes

Use force to drop a task and ignore errors if the task does not exist.

CREATE_TOOL Procedure

Use the DBMS_CLOUD_AI_AGENT.CREATE_TOOL procedure to register a custom tool that an agent can use during task processing. Each tool is identified by a unique tool_name and includes attributes that define its purpose, implementation logic, and metadata.

The tools can be custom PL/SQL procedures or functions, or built-in tool types such as RAG, SQL, WEBSEARCH, or NOTIFICATION.

Syntax

DBMS_CLOUD_AI_AGENT.CREATE_TOOL(
     tool_name          IN  VARCHAR2,
     attributes          IN  CLOB,
     status              IN  VARCHAR2  DEFAULT NULL,
     description         IN  CLOB      DEFAULT NULL
  );

Parameters

Parameter	Description
tool_name	A unique name to identify the tool. This is a mandatory parameter.
attributes	Tool attributes in JSON format. See CREATE_TOOL Attributes. This is a mandatory parameter.
status	Status of the tool. The default value is ENABLED.
description	User-defined description to help identify the tool. This value is stored in the database but not sent to the LLM. The default value is NULL.

Example

The following example creates a tool for to generate an email to the recipient.

BEGIN
  DBMS_CLOUD_AI_AGENT.CREATE_TOOL(
    tool_name  => 'NOTIFICATION',
    attributes => '{"tool_type": "EMAIL"}'
  );
END;
/

CREATE_TOOL Attributes

These attributes provide tool configuration using a JSON object in the attributes parameter. Each attribute configures how the agent uses the tool during reasoning.

Attributes

Attribute Name	Default Value	Description
instruction	-	A clear, concise statement that describes what the tool should accomplish and how to do it. This text is included in the prompt sent to the LLM. This is a mandatory parameter.
function	-	Specifies the PL/SQL procedure or function to call when the tool is used. This is a mandatory parameter for custom tools.
tool_type	-	Specifies a built-in tool type. If set, instruction and function are not required. The possible values are: SQL RAG WEBSEARCH NOTIFICATION You can create built-in tools with customized names and parameters. You can also create multiple built-in tools with same tool_type but different tool_params.
tool_params	-	Specifies parameters for registering built-in tools.
tool_inputs	-	Describes input arguments. Similar to column comments in a table. For example: "tool_inputs": [ { "name": "data_guard", "description": "Only supported values are "Enabled" and "Disabled"" } ]

Prerequisites for Built-In Tool

Perform these steps before you use built-in tools:

RAG Tool

1. Create an AI profile that references a vector index. See Select AI with Retrieval Augmented Generation (RAG).

NOTIFICATION Tool

The following are supported: SLACK and EMAIL.

SLACK Tool:

1. Set up a Slack application to receive messages from Autonomous Database.
2. Create a credential using DBMS_CLOUD_NOTIFICATION to send Slack notifications from Autonomous Database. See Send Slack Notifications from Autonomous AI Database.

EMAIL Tool:

Note:

The email tool sends email notification from OCI.

1. Identify your SMTP endpoint. For example, select one of the SMTP endpoints:

   smtp.us-phoenix-1.oraclecloud.com
   smtp.us-ashburn-1.oraclecloud.com
   smtp.email.uk-london-1.oci.oraclecloud.com
   smtp.email.eu-frankfurt-1.oci.oraclecloud.com

2. Grant SMTP access using an ACL. See Configuring SMTP Connection.
3. Generate SMTP credentials for email delivery. The UTL_SMTP package uses credentials to authenticate with email delivery servers when you send an email. See Generate SMTP Credentials for a User for more information.
4. Create approved senders. Complete this step for all email addresses you use as the From address with UTL_SMTP.MAIL subprogram. See Managing Approved Senders for more information.
5. Grant SMTP access to the user by adding Access Control Entry (ACE). For more details, see Send Email from Autonomous Database Using DBMS_CLOUD_NOTIFICATION.

HTTP Tool:

Grant network access using ACL to the user-specified endpoint to enable http request:

BEGIN
  DBMS_NETWORK_ACL_ADMIN.APPEND_HOST_ACE(
    host => 'www.example.com',
    ace  => xs$ace_type(
             privilege_list => xs$name_list('http'),
             principal_name => 'ADB_USER',
             principal_type => xs_acl.ptype_db
           )
  );
END;
/

Built-In Tool Types in DBMS_CLOUD_AI_AGENT Package

You can register built-in tools using the tool_type and tool_params attributes.

Tool Type	Description	Mandatory Parameters
SQL	Translates natural language into SQL queries.	profile_name
RAG	Performs Retrieval-Augmented Generation through vector search.	profile_name
NOTIFICATION	Sends messages to a Slack channel or sends emails using SMTP. Available options: slack email	notification_type. When notification_type is slack then credential_name and slack_channel are mandatory. When notification_type is email then credential_name, recipient, sender, and smtp_host are mandatory.
WEBSEARCH	Searches information from the web.	credential_name

ENABLE_TOOL Procedure

Use the DBMS_CLOUD_AI_AGENT.ENABLE_TOOL procedure to enable a previously registered tool. This changes the tool's status to ENABLED, allowing the agent to use it tasks.

Syntax

DBMS_CLOUD_AI_AGENT.ENABLE_TOOL(
   tool_name           IN  VARCHAR2
);

Parameters

Parameter	Description
tool_name	Name for the tool to enable. This parameter is mandatory.

Example

BEGIN
     DBMS_CLOUD_AI_AGENT.ENABLE_TOOL(
         tool_name    => 'SQL'
     );
END;
/

DISABLE_TOOL Procedure

Use the DBMS_CLOUD_AI_AGENT.DISABLE_TOOL procedure to disable a registered tool. This changes the tool's status to DISABLED, preventing the agent from using it in tasks.

Syntax

DBMS_CLOUD_AI_AGENT.DISABLE_TOOL(
   tool_name           IN  VARCHAR2
);

Parameters

Parameter	Description
tool_name	Name of the tool to disable. This parameter is mandatory.

Example

The following example disables an existing tool.

BEGIN
     DBMS_CLOUD_AI_AGENT.DISABLE_TOOL(
         tool_name    => 'SQL'
     );
END;
/

DROP_TOOL Procedure

Use the DBMS_CLOUD_AI_AGENT.DROP_TOOL procedure to remove a previously registered tool with the agent. If the task does not exist, you can suppress the error by setting the force parameter to TRUE.

Syntax

DBMS_CLOUD_AI_AGENT.DROP_TOOL(
   tool_name IN  VARCHAR2,
   force     IN  BOOLEAN DEFAULT FALSE
);

Parameters

Parameter	Description
tool_name	Name of the tool to drop. This is a mandatory parameter.
force	If TRUE, then the procedure ignores errors if tool does not exist. The default value for this parameter is FALSE.

Example

BEGIN
     DBMS_CLOUD_AI_AGENT.DROP_TOOL(tool_name => 'SQL');
END;
/

Usage Notes

Use force to drop a tool and ignore errors if the tool does not exist.

CREATE_TEAM Procedure

Use the DBMS_CLOUD_AI_AGENT.CREATE_TEAM procedure to define a team of AI agents that work together to accomplish tasks. You specify agents and their associated tasks, along with how the tasks should be processed.

Syntax

DBMS_CLOUD_AI_AGENT.CREATE_TEAM(
   team_name    IN VARCHAR2,
   attributes   IN CLOB,
   status       IN VARCHAR2 DEFAULT NULL,
   description  IN CLOB     DEFAULT NULL
);

Parameters

Parameter	Description
team_name	A unique name to identify the AI agent team. This is a mandatory parameter.
attributes	JSON-formatted string defining team attributes such as agents and process. See CREATE_TEAM Attributes. This is a mandatory parameter.
status	Status of the team. The default value is ENABLED.
description	User-defined description to identify the team's purpose. This value is stored in the database but not sent to the LLM. The default value is NULL.

Example

The following example creates an agent team to process a customer's product return.

BEGIN                                                                
  DBMS_CLOUD_AI_AGENT.CREATE_TEAM( 
    team_name  => 'ReturnAgency',                                                            
    attributes => '{"agents": [{"name":"Customer_Return_Agent","task" : "Return_And_Price_Match"}],
                    "process": "sequential"}');                                                                
END;                                                                      
/

CREATE_TEAM Attributes

These attributes provide AI agent team configuration. Each team has a unique team_name and uses a set of attributes to define its structure and execution strategy.

Attributes

Attribute Name	Default Value	Description
agents	-	Specifies JSON array listing agents and their corresponding tasks. An agent can be assigned to multiple tasks using the same name. This is a mandatory parameter. Example: [ {"name": "<agent_name>", "task": "<task_name>" ]
process	-	Specifies how tasks are run. Use "sequential" for tasks to run one after another in order.

SET_TEAM Function

Use DBMS_CLOUD_AI_AGENT.SET_TEAM to specify the AI agent team for the current session.

Once a team is set, only the select ai agent action is enabled, the system does not support other Select AI actions.

Syntax

DBMS_CLOUD_AI_AGENT.SET_TEAM(
    team_name      IN  VARCHAR2,
);

Parameters

Parameter	Description
team_name	Specifies the name of the AI Agent team to use in the current session. The team must already exist in the system. This parameter is mandatory.

Example

The following example sets the user-specified AI agent team in the current session.

EXEC DBMS_CLOUD_AI_AGENT.SET_TEAM('ReturnAgency');
  

GET_TEAM Function

Use DBMS_CLOUD_AI_AGENT.GET_TEAM to return the AI agent team for the current session.

Syntax

DBMS_CLOUD_AI_AGENT.GET_TEAM RETURN VARCHAR2;

Example

The following example returns the AI agent team in the current session.

SELECT DBMS_CLOUD_AI_AGENT.GET_TEAM from DUAL;
  

CLEAR_TEAM Function

Use DBMS_CLOUD_AI_AGENT.CLEAR_TEAM to clear the AI agent team set in the current session. This procedure does not drop the team.

Syntax

DBMS_CLOUD_AI_AGENT.CLEAR_TEAM;

Example

The following example clears the AI agent team in the current session.

BEGIN
  DBMS_CLOUD_AI_AGENT.CLEAR_TEAM;
END;
/
  

RUN_TEAM Function

Use DBMS_CLOUD_AI_AGENT.RUN_TEAM to start a new AI agent team or resume a paused one that is waiting for human input. If you provide an existing process ID and the associated team process is in the WAITING_FOR_HUMAN state, the function resumes the workflow using the input you provide as the human response.

Syntax

DBMS_CLOUD_AI_AGENT.RUN_TEAM(
     team_name               IN  VARCHAR2,
     user_prompt             IN  CLOB DEFAULT NULL,
     params                  IN  CLOB DEFAULT NULL,
  );

Parameters

Parameter	Description
team_name	Specifies the name of the AI Agent team to use in the current session. The team must already exist in the system. This parameter is mandatory.
user_prompt	Specifies the user query or response provided to the task. If the task is in the RUNNING state, the input acts as a placeholder for the {query} in the task instruction. If the task is in the WAITING_FOR_HUMAN state, the input serves as the human response.
params	Specifies customized parameters used during AI agent team process. Supported parameters include: conversation_id: Identifies the conversation session associated with the agent team. Use this parameter for stateless application integrations (for example, APEX) to distinguish between different user connections. variables: Define key-value pairs that provide additional input to the agent team. These variables help customize the team’s behavior during the process.

Example

The following example runs the user-specified AI agent team in the current session.

l_final_answer := DBMS_CLOUD_AI_AGENT.RUN_TEAM(
  team_name => 'ReturnAgency',
  user_prompt => 'I want to return computer desk chair',
  params => '{"conversation_id": "' ||l_conversation_id || '"}'
);

  

ENABLE_TEAM Procedure

Use the DBMS_CLOUD_AI_AGENT.ENABLE_TEAM procedure to enable a user-specified AI agent team. This changes the team's status to ENABLED.

Syntax

DBMS_CLOUD_AI_AGENT.ENABLE_TEAM(
   team_name           IN  VARCHAR2
);

Parameters

Parameter	Description
team_name	Name for the AI agent team to enable. This parameter is mandatory.

Example

The following example enables a user-specified AI agent team.

BEGIN
     DBMS_CLOUD_AI_AGENT.ENABLE_TEAM(
         team_name    => 'ReturnAgency'
     );
END;
/

DISABLE_TEAM Procedure

Use the DBMS_CLOUD_AI_AGENT.DISABLE_TEAM procedure to disable an existing AI agent team. This changes the team's status to DISABLED.

Syntax

DBMS_CLOUD_AI_AGENT.DISABLE_TEAM(
   team_name           IN  VARCHAR2
);

Parameters

Parameter	Description
team_name	Name of the AI agent team to disable. This parameter is mandatory.

Example

The following example disables an existing AI agent team.

BEGIN
     DBMS_CLOUD_AI_AGENT.DISABLE_TEAM(
         team_name    => 'ReturnAgency'
     );
END;
/

DROP_TEAM Procedure

Use the DBMS_CLOUD_AI_AGENT.DROP_TEAM procedure to remove a previously created AI agent team.

Syntax

DBMS_CLOUD_AI_AGENT.DROP_TEAM(
   team_name IN  VARCHAR2,
   force     IN  BOOLEAN DEFAULT FALSE
);

Parameters

Parameter	Description
team_name	Name of the team to drop. This is a mandatory parameter.
force	If TRUE, then the procedure ignores errors if team does not exist. The default value for this parameter is FALSE.

Example

BEGIN
     DBMS_CLOUD_AI_AGENT.DROP_TEAM(team_name => 'ReturnAgency');
END;
/

Usage Notes

Use force to drop a team and ignore errors if the team does not exist.