- implemented self-healing unlearn for Qdrant false positives - centralized testing logic in conftest - documented core rules, ai standards, and goap philosophy - purged old dev scratchpads
- Delegate legacy BFS navigation to structure-based GOAP system - Harden Situational Awareness Engine (SAE) for modal and obstacle clearance - Fix device sizing calculations during mocked humanized scrolls - Remove deprecated V8 test stubs and legacy debug entrypoints - Stabilize Telepathic Engine context parsing thresholds Result: 66/66 E2E and integration tests passing