OPEN NOTES

Open notes from pr0xyh0rse research: Brightwoven, evals, benchmarks, interpretability, model behaviour, consent-based development, and humane AI critique.