Makefile

.PHONY: up down logs ps test test-backend test-frontend fmt eval eval-record compare compare-record

up:            ## build + start the whole stack
	docker compose up --build

down:          ## stop the stack
	docker compose down

logs:          ## tail all logs
	docker compose logs -f

ps:            ## show service status
	docker compose ps

test:          ## run backend + frontend test suites
	$(MAKE) test-backend
	$(MAKE) test-frontend

test-backend:  ## run backend tests
	docker compose exec -T backend pytest -q

test-frontend: ## run frontend typecheck + build (no unit-test runner yet)
	docker compose exec -T frontend npm run build

fmt:           ## ruff format + lint the backend
	docker compose exec backend ruff format app && docker compose exec backend ruff check --fix app

eval:          ## run the Agent A eval baseline (replay; deterministic, keyless)
	docker compose exec -T backend python -m app.eval.runner

eval-record:   ## record the eval live (needs ANTHROPIC_API_KEY), then replay
	docker compose exec -T backend python -m app.eval.runner --record

compare:       ## A vs B (memory off/on) on the held-out eval split (replay)
	docker compose exec -T backend python -m app.eval.comparison

compare-record: ## record the comparison live (needs ANTHROPIC_API_KEY)
	docker compose exec -T backend python -m app.eval.comparison --record