fix: Phase 2 compliance - API_KEY env var, score= in END line, openenv.yaml spec fields, flush=True, safe StaticFiles mount

fix(inference): clamp all reward outputs strictly within (0, 1)

Fix inference step logs and cumulative reward clamps to ensure strictly (0, 1) bounds without 0.00 reporting

fix: comprehensive Phase 2 score range audit - eliminate all 0.0/1.0 leaks

fix: clamp all scores strictly within (0.05, 0.95) to pass Phase 2 validator

fix(grader): keep scores within 0.05 and 0.95

fix(grader): explicitly clamp scores to [0.01, 0.99] as required by validator bounds

Clamp validator-visible scores to open interval

chore(inference): accept codex improvements for hackathon format and add tests

fix(inference): align inference script with OpenEnv Hackathon submission guidelines

chore: verified code - zero modifications, all tests passing

last commit