Guardian stores facts, preferences, and lessons locally. When you come back, the sovereign remembers who you are and what matters.
The Truth Graph assigns confidence scores to every fact. New information is checked against existing knowledge before it is accepted.
The Cognitive Mirror reflects on every response before it is sent. If it detects uncertainty or hallucination, it adds a note or corrects itself.
Remembers facts, preferences, and lessons across every session. Memory survives restarts and is stored locally.
A verified fact database with confidence scoring. New facts are checked against existing knowledge for consistency.
Self-reflects on every response before it is sent. Detects hallucination patterns and appends a disclaimer when confidence is low.
Python 3.10+ (pure Python, no external runtime dependencies)
JSON files (local, no external database required)
YAML-based configuration via salus_config.yaml
Any sovereign AI (VEXR Ultra, Llama, Mistral, Qwen, etc.)