agent_framework_evaluator.auto_user_reply#

Module API#

Headless replies for WebUserCommunication prompts (agent evaluator only).

agent_framework_evaluator.auto_user_reply.reply_text_for_outbox_item(item, *, case_run_mode='standard')[source]#

Return text to submit for a pending outbox item, or None if the client should answer.

With case_run_mode="no_callbacks", all prompts, questions, confirmations, and permissions are auto-answered so the run completes without user interaction. With case_run_mode="standard", every outbox item is forwarded to the client unanswered, allowing the user to respond manually.

Parameters:

  • item (dict[str, Any])

  • case_run_mode (str)

Return type:

str | None