configure_detection
Set up loop detection parameters and progress metrics for AI agents to identify repetitive or non-progressing reasoning patterns, enabling intelligent recovery and task optimization.
Instructions
Configure loop detection parameters and domain-specific progress indicators
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| alternating_threshold | No | Threshold for detecting alternating action patterns (0.0-1.0) | |
| min_actions_for_detection | No | Minimum number of actions required before loop detection | |
| progress_indicators | No | Action patterns that indicate positive task progress (e.g., ["success", "complete", "found"]) | |
| progress_threshold_adjustment | No | How much to increase thresholds when progress indicators are present | |
| repetition_threshold | No | Threshold for detecting repetitive action patterns (0.0-1.0) | |
| semantic_intents | No | Domain-specific action intents for semantic analysis (e.g., ["navigating", "clicking", "typing"]) |