mindesbunister
|
c13cbaec4f
|
docs: Complete risk management bug analysis for new agent
CRITICAL: Four interconnected bugs causing ,000+ losses
- Bug #76: placeExitOrders() returns 2 sigs instead of 3 (SL missing)
- Bug #77: Position Manager logs 'monitoring' but isMonitoring=false
- Bug #78: Orphan cleanup cancelAllOrders() affects ALL positions
- Bug #80: Retry loop 'permanent fix' failed, still active
Latest incident (Dec 9, 21:45-21:56):
- Position: 6.15 SOL SHORT, +6.37 profit
- Orders: TP1/TP2 exist, slOrderTx NULL
- Health Monitor: 20+ CRITICAL alerts
- Outcome: Closed at TP2 +.69 (LUCKY)
- Risk: Could have been massive loss
Document includes:
- Complete timeline of fourth incident
- Root cause analysis for all bugs
- Failed fix analysis (Bug #80 cooldown ineffective)
- Detection system capabilities
- Emergency response procedures
- Action plan for new agent (5 phases)
- Test suite reference (113 tests)
- Multi-chunk close recording bug
- Drift API verification mandate
User statement: 'risk management vanished again' (fourth time)
User expectation: Permanent fixes, not temporary patches
Priority: Stop pattern of emergency restarts + order re-placements
|
2025-12-09 23:09:20 +01:00 |
|