Fetching metadata from the HF Docker repository... 
Update README.md
b97e2f2 verified - assets Pre-bake backbone card art (instant, no shimmer) + stop warming face-down enemy hand
- tests Pre-bake backbone card art (instant, no shimmer) + stop warming face-down enemy hand
- 84 Bytes Initial commit
- 4.67 kB Build Tabras card duel prototype
- 4.44 kB Add field notes (lessons learned) to README and FIELD_NOTES.md
- 9.35 kB Update README.md
- 3.29 kB Build Tabras card duel prototype
- 2.29 kB Trim prefetch to 1 pack, log model-pack failures, JSON-constrained cards, rotate fallback names
- 48 kB Polish demo: boss reveal screen, portraits, varied packs, paced draft, title copy
- 1.65 kB Hardcode Modal endpoint URLs in app_hf.py (public; env still overrides)
- 6.53 kB Add MODE switch: LOCAL (on-device, default) vs MODAL (GPU endpoints)
- 4.88 kB Build Tabras card duel prototype
- 5.38 kB finished base
- 7.83 kB Fix card truncation: lead JSON with effects+name, request concise flavor/art, bump card tokens
- 4.17 kB Pre-bake backbone card art (instant, no shimmer) + stop warming face-down enemy hand
- 1.96 kB finished base
- 7.64 kB Build Tabras card duel prototype
- 41.8 kB Polish demo: boss reveal screen, portraits, varied packs, paced draft, title copy
- 567 Bytes Build Tabras card duel prototype
- 12.2 kB Trim prefetch to 1 pack, log model-pack failures, JSON-constrained cards, rotate fallback names
- 5.24 kB Size Modal pools under the 10-GPU plan cap (card 3/4, art 3/4, boss 1/2)
- 15.5 kB Build Tabras card duel prototype
- 4.76 kB Build Tabras card duel prototype
- 277 Bytes finished base
- 15 Bytes Modal-backed Space: run models on Modal GPU endpoints, Space is a thin HTTP client
- 48.8 kB Warm boss card art in battle (GPU free post-draft) so boss plays show art instantly
- 52 Bytes Build Tabras card duel prototype
- 1.05 kB Integrate ZeroGPU Space path with card-cache, sampling, and diversity fixes
