honcho

Cross-session user modeling is maintained by configuring Honcho memory for Hermes, which learns user identity across conversations and provides a personalized experience. Use `hermes honcho setup` to initialize the memory system.

Multi-profile peer isolation is achieved by assigning each Hermes profile its own Honcho peer. This isolates profile-specific data while maintaining a unified, shared view of the user across all sessions.

Dialectic reasoning refines conclusions by performing iterative rounds of reasoning within the Honcho memory system. This process systematically challenges and improves initial answers before delivering a final response.

Context budget enforcement limits the size of context injected into the system by configuring Honcho memory boundaries. This prevents oversized session summaries and historical data from exceeding system constraints.

Observation configuration controls exactly what Honcho learns from each peer session. You define the observation parameters during setup to manage data collection and user modeling behavior.

Yes, the honcho-ai package is a required dependency for operation. The Skill configures and manages Honcho memory for Hermes but relies entirely on honcho-ai to execute the underlying memory functions.