Context Engine is parked, on purpose.

Agent memory — episodic, semantic, procedural — is a layer of AgentPact we refuse to fake. The previous version of this page claimed sub-5ms writes, four-nines SLAs, and SOC 2 compliance for a product with no backend. All of it is deleted.

The honest state: Context Engine sits at a decision gate. When the AgentPact feed layer produces a month of real usage data, it either becomes a thin, real MVP — or folds into NeuralForge, where IDC's memory governance work actually ships today:

• neuralforge-memory-mastery — the memory governance layer
• forge-recall — audit & verify existing memory
• forge-genesis — bootstrap a governed 6-layer system
• forge-hivemind — cross-agent memory consensus