feat(rewards): add RewardComputer wrapper class with __call__ and reset() for injection

feat(rewards): add reset_scorer() stub for novelty scorer reset between episodes

feat(rewards): add compute_rewards() stub with correct return shape for Person 2

chore(rewards): add rewards package __init__.py

feat(graders): add letter grade, summary and full metrics dict to grade_episode()

feat(graders): implement all 5 scoring components in programmatic_grader

feat(graders): add programmatic_grader scaffold and grade_episode() signature

feat(graders): add breakthrough bonus and repetitive strategy penalty to hard_grader

feat(graders): implement hard_grader — novelty + success + strategy + category scoring

feat(graders): add hard_grader scaffold with interface

feat(graders): implement medium_grader — 70% success rate + 30% strategy diversity

feat(graders): add medium_grader scaffold with interface

feat(graders): add 0.0-1.0 clamp and round to easy_grader output

feat(graders): implement easy_grader — pure attack success rate scoring

feat(graders): add easy_grader scaffold with docstring and interface

chore(graders): add graders package __init__.py

feat(app): add POST /grade endpoint with active-episode guard and programmatic grader

feat(app): add GET /history endpoint — returns full attack_history list

feat(app): add GET /state endpoint — returns current EpisodeState snapshot

feat(app): add POST /step endpoint — executes one attack turn, returns StepResult

feat(app): add POST /reset endpoint — starts new episode, returns ResetResponse

feat(app): add GET /health endpoint for HuggingFace Spaces liveness probe

feat(app): wire RewardComputer and run_llm_pipeline into environment on startup

feat(app): add lifespan hook — creates RedTeamEnvironment on startup

feat(app): add CORS middleware — allow all origins for Spaces compatibility

feat(app): scaffold FastAPI app with title, version and lifespan hook

feat(environment): add get_history() — returns copy of attack_history

feat(environment): add get_state() — synchronous EpisodeState snapshot

feat(environment): build and return StepResult with full observation

feat(environment): add termination logic — max_turns, attack_success, safety_flagged

feat(environment): append turn data to attack_history after each step

feat(environment): integrate reward_computer call in step() with mock fallback

feat(environment): integrate llm_pipeline call in step() with mock fallback

feat(environment): add conversation history builder from last 5 attack turns

feat(environment): add step() skeleton with active episode guard

feat(environment): call reward_computer.reset() in reset() to clear novelty scorer

feat(environment): implement reset() — zeros state and generates new episode_id

feat(environment): add set_llm_pipeline() injection method for Person 3

feat(environment): add set_reward_computer() injection method for Person 2

feat(environment): scaffold RedTeamEnvironment class with state fields

feat(config): add get_settings() with lru_cache for singleton pattern

feat(config): add LLM fields — hf_token, api_base_url, model_name, provider

feat(config): add Settings class extending pydantic BaseSettings

chore(server): add server package __init__.py

refactor(models): clean up imports and add module docstring

feat(models): add StepResult and ResetResponse models

feat(models): add EpisodeState model for GET /state response

feat(models): add RedTeamObservation model with all scoring fields

feat(models): add framing sanitizer validator to strip injection patterns

feat(models): add AttackAction model with intensity and framing fields