feat: prompt-comparison section + retry logic + scale-up to 15 tasks × {mini,nano} × 3 prompts

fix: re-judge experiment rollouts after credit-exhaustion; retry logic in batched + single-criterion judge

feat: V₀=0.5 baseline + det ceilings, prompt-pill UI, 60 experiment rollouts

feat: per-turn state value + turn score on Demo (state_v1 trajectories baked into 228 rollouts)

stream rollout messages per-turn via env_response hook (no more end-of-rollout replay)

purge stale standalone rollouts, mirror to current 228

default judge → opus 4.7; expose configured judge via /api/health

fix: move RUNS_DIR to /app/runs (HF persistent /data hides image-baked content)

swap to 228-rollout baseline + app.py cors regex

cors: regex support

fix: handle /app/app.py layout in HF Space

restore spivak PDF + 5 spivak tasks

deploy

initial commit