title: OpenAI web_search minimal reasoning gate

scenario:
  id: openai-web-search-minimal
  surface: model-provider
  coverage:
    secondary:
      - runtime.reasoning-and-cache-controls
      - web-search.openai-native-web-search
      - tools.web-search
  objective: Verify the OpenAI web_search minimal-reasoning E2E client distinguishes successful grounded turns from provider schema rejection.
  successCriteria:
    - Reject mode accepts the expected raw OpenAI schema rejection and the gateway schema wrapper.
    - Reject mode fails if the agent run unexpectedly succeeds or fails for unrelated transport reasons.
    - Success mode requires an `ok` agent result with the expected marker in visible reply payloads.
    - Gateway ports are parsed strictly before connecting.
  docsRefs:
    - docs/tools/web.md
    - docs/help/testing.md
    - docs/concepts/qa-e2e-automation.md
  codeRefs:
    - scripts/e2e/lib/openai-web-search-minimal/client.mjs
    - scripts/e2e/openai-web-search-minimal-docker.sh
    - test/e2e/qa-lab/runtime/docker-e2e-lane.ts
    - test/e2e/qa-lab/runtime/openai-web-search-minimal.e2e.test.ts
  execution:
    kind: script
    path: test/e2e/qa-lab/runtime/docker-e2e-lane.ts
    summary: Runs the OpenAI web_search Docker E2E lane through a real Gateway with a container-local OpenAI Responses mock.
    args:
      - --lane
      - openai-web-search-minimal