nonstop

To keep an AI agent working autonomously after you go AFK, you arm a stop-hook that intercepts turn termination events and nudges the agent to continue executing the task list. This enforces continuous progress across multi-step phases like planning and implementation.

A stop-hook loop prevents premature stopping by catching the agent's end-of-turn signal and issuing a nudge event instead of terminating. This guardrailed mechanism resolves ambiguity early and applies fallback strategies for blocked operations, keeping the workflow active.

Yes, you can safely arm and disarm autonomous workflow execution using local hook flag files. These files control the stop-hook state, allowing you to activate nonstop mode, check execution status, and safely deactivate the loop when unattended work is complete.

When an autonomous agent encounters a blocked or failing operation during nonstop execution, the guardrailed decision loop enforces safe fallback strategies. It prioritizes early ambiguity resolution and uses background handling to manage long-running operations without terminating the turn.

Yes, safe pre-flight validation is required before starting hands-free task list execution. This validation checks the environment and task parameters prior to arming the stop-hook, ensuring the autonomous workflow can proceed securely without immediate failures.

To check the progress of an AFK mode workflow, you use the status control command to query the active stop-hook state. This provides visibility into the current execution phase, allowing you to monitor unattended work without interrupting the autonomous loop.