An Agentforce created a custom Agent action, but it is not being picked up by the planner service in the
correct order.
Which adjustment should the Al Specialist make in the custom Agent action instructions for the planner service to work as expected?
A. Specify the dependent actions with the reference to the action API name.
B. Specify the profiles or custom permissions allowed to invoke the action.
C. Specify the LLM model provider and version to be used to invoke the action.
A. Specify the dependent actions with the reference to the action API name.
Explanation:
In AgentForce, the Planner Service determines the execution order of multiple custom actions by analyzing their dependencies. Here’s how to configure it:
Dependency Specification
In the custom action’s configuration, explicitly reference the API names of dependent actions.
Example: If Action_B requires output from Action_A, specify Action_A as a dependency in Action_B’s setup.
This ensures the Planner Service sequences actions logically (e.g., Action_A → Action_B → Action_C).
Why Not the Other Options?
B. Profiles/Permissions:
Controls who can run actions, not when they run.
C. LLM Model Provider:
Defines the AI model used for processing, not action sequencing.
Key Benefit:
Avoids errors like race conditions or missing data by enforcing step-by-step execution.