agent

Usage

agent: agent_name {
  advanced_analytics: yes | no
  description: "string"
  instructions: "string"
  is_dashboard_agent: yes | no
  show_debuginfo: yes | no
  show_thinking: yes | no
}
Hierarchy
Model file or LookML file
agent
Default value
None

Accepts
  • advanced_analytics: yes | no (Defaults to no).
  • description: A string describing the agent.
  • instructions: A string specifying custom prompts and behavior rules. Supports LookML constants (@{CONSTANT_NAME}).
  • is_dashboard_agent: yes | no (Required as yes for dashboard agents).
  • show_debuginfo: yes | no (Defaults to no).
  • show_thinking: yes | no (Defaults to no).

Definition

The agent parameter declares a data agent in LookML. With dashboard agents, developers can define custom agent instructions, analytical behavior, and display preferences in version-controlled LookML and associate the agent with LookML dashboards using the default_dashboard_agent or dashboard_agents parameters.

For detailed information on configuring and using LookML dashboard agents, see the Create and manage LookML dashboard data agents documentation page.

Subparameters

Dashboard agent definitions support the following subparameters:

advanced_analytics

The advanced_analytics subparameter controls whether the agent can use Advanced Analytics to generate and run Python code for complex computations and custom visualizations.

Accepts yes or no. The default value is no.

description

The description subparameter provides an optional text description of the agent's purpose.

agent: sales_agent {
  description: "Executive sales overview dashboard agent"
  is_dashboard_agent: yes
}

instructions

The instructions subparameter accepts a free-form string that provides the agent with business context, preferred fields, filtering rules, or tone of voice when processing conversational questions.

agent: sales_agent {
  instructions: "Focus on net revenue and filter by current quarter by default."
  is_dashboard_agent: yes
}

You can reference LookML constants within instructions by using @{<var>CONSTANT_NAME</var>} syntax:

agent: sales_agent {
  instructions: "Focus on net revenue. @{FISCAL_YEAR_RULE}"
  is_dashboard_agent: yes
}

If omitted, the agent operates without custom instruction context.

is_dashboard_agent

The is_dashboard_agent subparameter specifies whether the agent is configured as a dashboard agent.

Accepts yes or no. To use the agent with a LookML dashboard, set this parameter to yes.

show_debuginfo

The show_debuginfo subparameter determines whether the agent displays execution logs and diagnostic information in conversation responses.

Accepts yes or no. The default value is no.

show_thinking

The show_thinking subparameter determines whether the agent displays its reasoning steps and intermediate query planning in conversation responses.

Accepts yes or no. The default value is no.

Examples

Defining a LookML dashboard agent

The following example defines a LookML dashboard agent with custom instructions, Advanced Analytics enabled, and thinking visibility turned on:

agent: customer_insights_agent {
  instructions: "Address the user politely. Prioritize active subscriber metrics and churn analysis."
  description: "Dashboard agent for Customer Insights dashboard."
  is_dashboard_agent: yes
  advanced_analytics: yes
  show_thinking: yes
  show_debuginfo: no
}

To use this agent on a LookML dashboard, map it in the .dashboard.lookml file:

- dashboard: customer_insights
  title: "Customer Insights"
  layout: newspaper
  enable_dashboard_agent: true
  default_dashboard_agent: customer_insights_agent
  dashboard_agents: [customer_insights_agent]

  elements:
  - name: active_subscribers
    type: single_value
    model: subscriptions
    explore: customers
    measures: [customers.count]