amcos-failure-detection

Remote agent failure detection during heartbeat checks identifies offline or slow agents by applying classification logic to heartbeat intervals, determining the failure type and guiding the appropriate recovery path.

Classifying unacknowledged message delivery failures involves applying detection logic to message timeouts, categorizing the failure type, and producing structured outputs for incident logging and escalation.

Tracking task progress timeouts to identify an unresponsive agent requires monitoring task execution intervals, applying failure-detection rules, and classifying the timeout to determine the correct recovery actions.

Failure detection integrates with the AI Maestro registry and agent-status queries by applying classification rules during heartbeat checks and message delivery to generate structured incident logs for escalation.

Triggering a failure-detection cycle against a target agent initiates the detection and classification process, evaluating heartbeat checks, message delivery, and task progress timeouts to determine the failure type.

Heartbeat checks for remote agent failure detection are limited to identifying offline or slow agents based on interval timeouts, and must be combined with message delivery and task progress tracking for full classification.