forked from cardosofelipe/pragma-stack
Felipe Cardoso
05b75de21f
Add reflection layer for memory system with pattern detection, success/failure factor analysis, anomaly detection, and insights generation. Enables agents to learn from past experiences and identify optimization opportunities. Key components: - Pattern detection: recurring success/failure, action sequences, temporal, efficiency - Factor analysis: action, context, timing, resource, preceding state factors - Anomaly detection: unusual duration, token usage, failure rates, action patterns - Insight generation: optimization, warning, learning, recommendation, trend insights Also fixes pre-existing timezone issues in test_types.py (datetime.now() -> datetime.now(UTC)).