dkAmulet/sql-query-optimizer

fix: wrap TASK_GRADERS registry with _safe() — scores can never be 0.0 or 1.0

fix: force rebuild with clamped scores in all files

fix: _clamp() ensures all scores strictly in (0.001, 0.999)

fix: Pydantic field_validator clamps all scores to strictly (0.001,0.999) at model level

fix: clamp all reward scores to (0.001,0.999) at HTTP response level

fix: clamp all scores and breakdown values to strictly (0.001, 0.999)

fix: clamp scores to (0.001, 0.999) — validator requires strictly between 0 and 1

fix: use python:3.11-slim-bookworm to avoid stale digest, single-stage build

fix: robust inference.py with try/except, fallback queries, graceful HF_TOKEN handling

fix: server entry point must be main() function

fix: add missing server/app.py entry point

fix: resolve openai version conflict, add openenv-core>=0.2.0

fix: add pyproject.toml for openenv multi-mode deployment

fix: simplify openenv.yaml to pass openenv validate

fix: inference.py matches required START/STEP/END format and env var spec

fix: add HF Spaces YAML metadata to README

feat: SQL Query Optimizer OpenEnv v1.0