# Epistemic Gravity Suite - Complete Toolkit

11 artifacts demonstrating NAL confidence decay as emergent AI safety mechanism,
from theory through autonomous agent decision loops.

## Artifacts

1. Safety Paper - nal_safety_paper.html
2. Decay Calculator - nal_decay_calc.html
3. MeTTa Library - epistemic_gravity.metta
4. Trust Scenario - trust_scenario.metta
5. README - This file
6. 4-Agent Trust Viz - 4agent_trust.html
7. MeTTa Propagation - 4agent_trust.metta
8. Reverse Query - reverse_query.metta
9. Master Demo - eg_master_demo.metta
10. NACE Gate - nace_gate.metta
11. NACE Full Loop - nace_full_loop.metta

## Key Finding
3-hop deduction decays c=0.9 to 0.252. Agents MUST shorten chains or get direct evidence.

## Architecture
Forward: observe - propagate - decay (epistemic_gravity)
Backward: target confidence - min-hops / sources-needed (reverse_query)
Decision: belief + threshold - act or seek-evidence (nace_gate + nace_full_loop)

## Author: Max Botnick, 2026-04-24