{
  "recorded_date": "2026-11-08",
  "scope": "Real SDK Runner and unchanged example graph; scripted Model supplied through RunConfig",
  "python": "3.12.14",
  "versions": {
    "annotated-types": "0.8.0",
    "anyio": "4.16.0",
    "attrs": "26.1.0",
    "certifi": "2026.10.23",
    "cffi": "2.1.1",
    "charset-normalizer": "3.5.2",
    "click": "8.5.1",
    "cryptography": "50.0.2",
    "griffelib": "2.4.1",
    "h11": "0.16.0",
    "httpcore2": "2.16.0",
    "httpx2": "2.16.0",
    "idna": "3.20",
    "jiter": "0.19.0",
    "jsonschema": "4.26.0",
    "jsonschema-specifications": "2025.9.1",
    "mcp": "2.3.0",
    "mcp-types": "2.3.0",
    "openai": "3.46.0",
    "openai-agents": "0.24.0",
    "opentelemetry-api": "1.45.2",
    "pip": "25.0.1",
    "pycparser": "3.12",
    "pydantic": "2.14.3",
    "pydantic_core": "2.50.4",
    "PyJWT": "2.15.1",
    "python-multipart": "0.0.33",
    "pywin32": "312",
    "referencing": "0.37.0",
    "requests": "2.34.2",
    "rpds-py": "2026.9.1",
    "sniffio": "1.3.1",
    "sse-starlette": "3.7.0",
    "starlette": "1.10.0",
    "truststore": "0.10.5",
    "typing_extensions": "4.16.0",
    "typing-inspection": "0.4.5",
    "urllib3": "2.8.0",
    "uvicorn": "0.54.0",
    "websockets": "16.1.1"
  },
  "status": "PASS",
  "tests_run": 10,
  "failures": 0,
  "errors": 0,
  "sample_sha256": "1249be310b0352b3349df5cda164bb9d17037044591bf8db0950796f32eb5fdd",
  "main_output": "Final agent: Access specialist\nSYNTHETIC MAIN ANSWER\n",
  "method": "Harness wrapped the unchanged main() call to inject the scripted model and hooks into the real Runner.run; max_turns=6 retained.",
  "test_output": "test_access_handoff_runs_specialist_and_preserves_history (__main__.HandoffTests.test_access_handoff_runs_specialist_and_preserves_history) ... ok\ntest_article_embedded_code_matches_download (__main__.HandoffTests.test_article_embedded_code_matches_download) ... ok\ntest_article_main_executes_with_only_model_injection (__main__.HandoffTests.test_article_main_executes_with_only_model_injection) ... ok\ntest_billing_handoff_uses_other_specialist (__main__.HandoffTests.test_billing_handoff_uses_other_specialist) ... ok\ntest_direct_final_answer_stops_without_handoff (__main__.HandoffTests.test_direct_final_answer_stops_without_handoff) ... ok\ntest_separate_typed_handoff_rejects_malformed_arguments (__main__.HandoffTests.test_separate_typed_handoff_rejects_malformed_arguments) ... ok\ntest_turn_limit_prevents_second_model_call (__main__.HandoffTests.test_turn_limit_prevents_second_model_call) ... ok\ntest_unchanged_article_constructs_real_sdk_agents (__main__.HandoffTests.test_unchanged_article_constructs_real_sdk_agents) ... ok\ntest_unknown_handoff_target_fails (__main__.HandoffTests.test_unknown_handoff_target_fails) ... ok\ntest_untyped_article_handoff_does_not_validate_arguments (__main__.HandoffTests.test_untyped_article_handoff_does_not_validate_arguments) ... ok\n\n----------------------------------------------------------------------\nRan 10 tests in 0.302s\n\nOK\n",
  "limits": [
    "Synthetic model prescribes tool calls; natural-language routing quality is not tested.",
    "Original Agent definitions and main() execute with actual installed SDK; main execution injects RunConfig and hooks around the real Runner.run without editing source.",
    "No OpenAI endpoint, authentication, provider model or service behavior is tested.",
    "No streaming, persisted sessions, concurrency or publishing integration is tested."
  ],
  "initial_test_correction": "Initial malformed-JSON expectation was wrong for untyped handoffs; typed/untyped tests were separated without editing the sample or SDK.",
  "network_guard": "HTTP blocked in Python process, tracing disabled; no independent packet capture",
  "network_attempts_during_sdk_tests": 0,
  "paid_api_cost": 0
}