model-sovereigntylisted

# Model Sovereignty Model sovereignty is the practice of making deliberate decisions about which models to use, where they run, and whether to create custom models. It extends the framework's Theme #2 (Agency and Sovereignty) into the model layer. This skill guides practitioners through the decision framework from cross-cutting Theme #17 and Appendix P of the framework. ## The Decision Hierarchy Exhaust simpler approaches before escalating complexity. Each step adds maintenance burden. 1. **Prompting + context engineering** — the default. Most teams underestimate how far this carries them. Exhaust it first. 2. **RAG** — when the limitation is knowledge (volatile, large, or frequently changing information). 3. **Fine-tuning (LoRA/QLoRA)** — when the limitation is behaviour (consistent domain-specific patterns at scale). 4. **Distillation** — when the limitation is size or speed (edge deployment, latency-sensitive applications). 5. **Local hosting** — when the limitation is privacy, cost at scale, or independence from vendor defaults. ## The Decision Framework Walk through these questions in order. Stop at the first "yes." **Does your data require local processing?** PII, regulated data, trade secrets, or data subject to residency requirements → local hosting is non-negotiable for those interactions. Consult `references/decision-framework.md`. **Does knowledge change frequently?** Information changing weekly/monthly → add a RAG layer regardless of hostin