Failure Modes¶

For the complete list of all detectors with F1 scores, accuracy benchmarks, and detection methods, see the Detection Reference.

For the MAST taxonomy categories with detailed descriptions and real-world examples, see:

• Planning Failures (F1-F5)
• Execution Failures (F6-F11)
• Verification Failures (F12-F14)
• Extended Detectors (loops, injection, corruption, etc.)
• n8n · LangGraph · Dify · OpenClaw