Resonant Field Mapping (RFM) is a runtime control architecture for large language models that separates tone modulation, memory persistence, capability authority, and divergence stabilization into explicit governed layers.
This repository contains the v1.0 architecture paper describing a governed runtime control layer for LLMs, including:
- Resonant Field Modulation (RFM-M)
- Goal-Guarded Memory (GGM) with quarantine rules
- Governance Layer with deny-by-default scoped capability leases
- Divergence-triggered reversible degradation (SDB-1)
- Deterministic re-entry conditions
This specification focuses on bounded control over escalation, reversible actions under estimator disagreement, and constrained memory authority.
📄 Download the full paper (PDF)
Architecture specification; no reference implementation included.
This work is licensed under Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0). See the LICENSE file for details.