Fix grader import path: use root-level graders module instead of server.graders

Add score:0.5 to all YAML tasks, make graders callable without args, fix health status

Bypass cached validator endpoints

chore: force HF rebuild with deploy canary v3

chore: trigger rebuild

Iteration_4: expand task bank 20→30, add variants for all tasks, fix baseline policy for chain-gated CRIT tasks

feat: enforce investigation discipline + fix easy-task grading + add investigation_hints

feat: reproducible baseline — fixed seed, correct_action in table, accuracy summary

fix: per-task reward display uses weighted_reward key from grader

feat: run LLM inference baseline + fix SSL and loop guard in inference.py

docs: fix README gaps - complete action/observation tables and task descriptions

fix: add HF Space YAML frontmatter to README.md

feat: Iteration_2 — trajectory reward shaping, episode jitter, flag identification

chore: pre-iteration-1 snapshot — 75/75 e2e passing, baseline bug fixed

Initial Commit