{{agentToolTypeUiData.label}}
{{ agentToolTypeUiData.blueBoxDescription || agentToolTypeUiData.description}}.
The VertexAI vector store doesn't support document-level security.
To use this property, please select a Security tokens column in the Embedding Recipe {{rkFullInfo.creatingRecipe.name}} advanced properties.
Retrieval
If enabled, all metadata items of the document are retrieved and sent to the Agent
Sources
Filtering
If enabled, query level filtering will be supported
Saved Model not available. It has been disabled, retired, or is not supported for this feature. Choose another Saved Model or contact your administrator.
Channel settings
Variables
Variable {{ undefinedTemplateVariables[0] }} is used but not defined below. Add it to the list.
{{ undefinedTemplateVariables.length }} variables are used but not defined below. Add them to the list.
Variable {{ unusedToolVariables[0] }} is not used. Remove it from the list.
{{ unusedToolVariables.length }} variables are not used. Remove them from the list.
Available variables to customize the message to be sent (content, recipients etc.). Can be used in the template with ${variable} syntax.
Tool description
The endpoint {{uiState.persistedApiEndpointName}} has been deleted, or you don’t have the required permissions to access or query it through the deployer.
Only endpoints with an OpenAPI documentation are available
You can define the authorization to query an endpoint through the deployer in the settings of its deployment.
Parsed from the description of the operation defined in the OpenAPI documentation
The resulting description must clearly state the role of the endpoint. It should also specify the expected input, as well as how to interpret the output. Indicating the exact format of both input and output is helpful for correct integration and usage.
Example: The "Customer Success > Predict Churn" endpoint predicts the probability that a customer will churn, based on several criteria such as account age, recent activity level, number of support tickets, and subscription type. The output is a number between 0 (unlikely to churn) and 1 (very likely to churn).--some-key somevalue is 2 arguments
MCP tools
Note that the selected code environment must use Python 3.10 to 3.13. It must include the
Use the Agent tool MCP servers support container runtime addition to add
Use the Test feature to get more debug info.
fastmcp package.Use the Agent tool MCP servers support container runtime addition to add
uvx and npx to the code env container image.Use the Test feature to get more debug info.
{{agentToolTypeUiData.label}}
{{agentToolTypeUiData.description}}.
Advanced
Configure advanced Dataiku-specific properties for this tool. Usually used at the direction of Dataiku Support.