fix: safe int/float conversion for LLM action fields

fix: explicit step() payload, urllib.error import, no **action spread

fix: match sample format exactly — :.2f rewards, :.3f score, finally block

fix: use 0.001/0.999 bounds matching openenv.yaml score_range — app.py

fix: use 0.001/0.999 bounds matching openenv.yaml score_range — graders.py

fix: use 0.001/0.999 bounds matching openenv.yaml score_range — inference.py

fix: use 0.001/0.999 bounds matching openenv.yaml score_range — env.py

fix: clamp all scores strictly (0,1) — inference.py

fix: clamp all scores strictly (0,1) — env.py

fix: add WebSocket /ws endpoint and fix reset response structure

fix: clamp episode_score at API boundary

fix: clamp episode_score at API boundary

fix: NaN/inf safe_score

fix: NaN/inf safe_score

fix: accept full task IDs in reset and step endpoints

fix: compute score from avg rewards with 1e-6 clamp

fix: prevent randint crash when expiry bounds are equal or inverted

fix: correct env name to retail-inventory-expiry in START log line

fix: rewards list uses final episode_score for validator consistency

fix: use 3dp formatting so clamped rewards never print as 0.00 or 1.00

fix: rewrite stdout format to match required START/STEP/END spec

fix: remove dummy-key default from HF_TOKEN

fix: add reward field to Observation per OpenEnv standard

fix: add reward field to Observation per OpenEnv standard

fix: add /metadata /schema /mcp endpoints and fix /health status

fix: clamp reward total in step response

sync: upload all files with latest fixes

sync: upload all files with latest fixes

sync: upload all files with latest fixes

Upload env.py with huggingface_hub

Upload inference.py with huggingface_hub

Upload inference.py with huggingface_hub

Upload inference.py with huggingface_hub

Upload inference.py with huggingface_hub

Upload graders.py with huggingface_hub

Upload inference.py with huggingface_hub

Upload inference.py with huggingface_hub

Upload env.py with huggingface_hub

Upload simulation.py with huggingface_hub

Upload server/__init__.py with huggingface_hub

Upload server/app.py with huggingface_hub

Upload uv.lock with huggingface_hub

Upload pyproject.toml with huggingface_hub

Upload README.md with huggingface_hub

Upload folder using huggingface_hub

initial commit