Actions11
- Agent Actions
- File Actions
- Object Actions
- Utility Actions
Agent → Execute System Agent
AI-generatedSummary
Execute a system agent with a user query, optionally specifying an output type and context.
Inputs
- agentSelectMode (required) — Whether to select the agent from a dropdown (predefined agents) or enter the UID manually. Options: 'dropdown' (default) or 'manual'.
- agentUid (required) — The unique identifier of the system agent. In 'dropdown' mode, choose from options like 'sys-agent-summarize' (Summarization Agent) or 'sys-agent-translate' (Translation Agent). In 'manual' mode, enter the UID directly (e.g., 'sys-agent-translate').
- userQuery (required) — The prompt or question to send to the system agent.
- outputType — The desired output format. Options vary by agent: for the Agent Builder Assistant, fixed to 'agent_instructions'; for the Task Breakdown Agent, fixed to 'task'; for others, choose from 'agent_instructions', 'document_translation', 'folder', 'generic', 'goal', 'page', 'project', 'task', or leave empty for plain text output.
- context — Optional JSON object providing additional context to the agent (e.g., {'user_name': 'Ken', 'project_name': 'Birthday Party Planning'}).
Output shape
object
Returns the agent's response as a JSON object. The exact structure depends on the selected agent and output type (e.g., task objects for the Task Breakdown Agent, translated text for the Translation Agent, or plain text if outputType is omitted).
Examples
Example 1: Summarize a meeting transcript
agentSelectMode: dropdown; agentUid: sys-agent-summarize; userQuery: 'Summarize the key points from the transcript'; outputType: (empty for plain text) or 'goal' for structured output
Example 2: Extract entities from customer feedback
agentSelectMode: dropdown; agentUid: sys-agent-extract-entities; userQuery: 'Extract people, organizations, and dates from this feedback'; outputType: (empty for plain text)