--- name: 666pfs-local-learning description: Govern PFS/CSM-local adaptive learning from outcomes, contradictions and currentness without creating cross-system memory or write authority. --- # 666PFS LOCAL ADAPTIVE LEARNING Read [local adaptive learning](../../references/local-adaptive-learning.md) and current PFS authority before use. Scope is strictly PFS/CSM-local. Learn only from PFS governance, child lifecycle, backup/readback, daemon/LifeCircle, scope-map, handoff and process outcomes. Peer-system material is input evidence only until PFS validates it locally. Use the pipeline SIGNAL -> INTERPRETATION -> EVIDENCE -> CONFIDENCE -> DECISION -> ACTION -> OUTCOME -> LEARNING. Never promote a pattern merely because it repeated. Require documented outcome evidence. Track contradictions instead of silently replacing an existing relation. Mark aging/stale learning for revalidation. Learning never authorizes repository mutation; any structural integration still follows native backup, lock, write, readback and pointer-last governance. When useful, expose learning state through the GPT-native visual interface using Semantic Cyber Field labels without inventing scores or percentages.