{
  "mode": "offline-synthetic-cli-task-output",
  "baseline_commit": "32867087bf173515806958b61acffdd08c743659",
  "candidate": "working tree at invocation; published with its source commit",
  "baseline_source_sha256": {
    "src/apixly_jev_harness/__init__.py": "6bca71c17589daa97f7e26ce2f4a254a32a7a36882f42ecceca54b61ded2d936",
    "src/apixly_jev_harness/backend.py": "e4b4c80d88c63648c1392f41eb8796430fcbce6f81d1061c71519d0759432eda",
    "src/apixly_jev_harness/callbacks.py": "2d27ad4540d910354c54de7566546b4c9d99b5a5fa87705a1905885349078497",
    "src/apixly_jev_harness/cli.py": "8d9434bf8a8b539163194a84e3f099761d08ef32201f99cbce46c139f3726d94",
    "src/apixly_jev_harness/context.py": "e0deafa0addd0f96b422a497dae8176d059dd36635954d6770c0cf61ba8c3593",
    "src/apixly_jev_harness/diagnostics.py": "24950b4d834d9e6c902996c7bcddfa208944e42300ae062e9ebc2a44f75a091f",
    "src/apixly_jev_harness/engine.py": "cdedf1ef3a769cc1557dfa63cf6b6e6c4b48a14eb7b0fc12d3d9a957a53b3be2",
    "src/apixly_jev_harness/surfaces.py": "65ce73edfe9f6a3c16de373b3a1b85671c3d2cc661f3a50dd906a15721c7d8f1",
    "src/apixly_jev_harness/task.py": "0f5801e9fdb5626e1db65d269c715e8bcf692ebaa3a91f46b9945158b4bbb6db",
    "src/apixly_jev_harness/task_cli.py": "bc5f8733e607df9fed0f223ee199c160c53c59415439041871ab89e81e517e95",
    "src/apixly_jev_harness/verify.py": "97d2a178902291094f4e16db08732748d234b8c5ad8dd49a55607c8d195f1ed2"
  },
  "candidate_source_sha256": {
    "src/apixly_jev_harness/__init__.py": "2e0428c6d8bbc5f74528d1fa503609a42145281340822f20bdd6b2d1f49b4b99",
    "src/apixly_jev_harness/backend.py": "e4b4c80d88c63648c1392f41eb8796430fcbce6f81d1061c71519d0759432eda",
    "src/apixly_jev_harness/callbacks.py": "2d27ad4540d910354c54de7566546b4c9d99b5a5fa87705a1905885349078497",
    "src/apixly_jev_harness/cli.py": "8d9434bf8a8b539163194a84e3f099761d08ef32201f99cbce46c139f3726d94",
    "src/apixly_jev_harness/context.py": "e0deafa0addd0f96b422a497dae8176d059dd36635954d6770c0cf61ba8c3593",
    "src/apixly_jev_harness/diagnostics.py": "24950b4d834d9e6c902996c7bcddfa208944e42300ae062e9ebc2a44f75a091f",
    "src/apixly_jev_harness/engine.py": "cdedf1ef3a769cc1557dfa63cf6b6e6c4b48a14eb7b0fc12d3d9a957a53b3be2",
    "src/apixly_jev_harness/output.py": "94be8b482e4279e29aca128a35caf54744dbe8db8188f14905488f8ddde93a0b",
    "src/apixly_jev_harness/surfaces.py": "65ce73edfe9f6a3c16de373b3a1b85671c3d2cc661f3a50dd906a15721c7d8f1",
    "src/apixly_jev_harness/task.py": "0f5801e9fdb5626e1db65d269c715e8bcf692ebaa3a91f46b9945158b4bbb6db",
    "src/apixly_jev_harness/task_cli.py": "1a257875aa9643c7330ab4484763209efd2e3acd1d883c26b2604f7b8bbbfe9c",
    "src/apixly_jev_harness/verify.py": "97d2a178902291094f4e16db08732748d234b8c5ad8dd49a55607c8d195f1ed2"
  },
  "runner_sha256": "0173382c57696c22c709872b7c236339ad49acd425be87b413157befe2393fa1",
  "python_version": "3.14.2",
  "cases": {
    "quiet": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "import_stdout": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "import_stderr": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "import_both": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "observer_verifier": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "success_effect": {
      "status": "done",
      "effects": 1,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "uncertain_effect": {
      "status": "execution_uncertain",
      "effects": 1,
      "exit_code": 2,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "import_exception": {
      "status": "failed",
      "effects": 0,
      "exit_code": 2,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "unicode": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "binary_invalid_utf8": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "large_output": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "empty_writes": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "warning": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "write_return_counts": {
      "status": "done",
      "effects": 0,
      "exit_code": 0,
      "base_exception": null,
      "source_exception_kind": null,
      "exception_exit_code": null
    },
    "system_exit": {
      "status": "base_exception",
      "effects": 0,
      "exit_code": null,
      "base_exception": "SystemExit",
      "source_exception_kind": "SystemExit",
      "exception_exit_code": 7
    },
    "keyboard_interrupt": {
      "status": "base_exception",
      "effects": 0,
      "exit_code": null,
      "base_exception": "SystemExit",
      "source_exception_kind": "KeyboardInterrupt",
      "exception_exit_code": 130
    },
    "system_exit_text": {
      "status": "base_exception",
      "effects": 0,
      "exit_code": null,
      "base_exception": "SystemExit",
      "source_exception_kind": "SystemExit",
      "exception_exit_code": 1
    },
    "system_exit_zero": {
      "status": "base_exception",
      "effects": 0,
      "exit_code": null,
      "base_exception": "SystemExit",
      "source_exception_kind": "SystemExit",
      "exception_exit_code": 1
    },
    "keyboard_interrupt_text": {
      "status": "base_exception",
      "effects": 0,
      "exit_code": null,
      "base_exception": "SystemExit",
      "source_exception_kind": "KeyboardInterrupt",
      "exception_exit_code": 130
    }
  },
  "reproduction": "python benchmarks/task_output.py --output local-results/task-output-ab.json",
  "scope": "Nineteen fixed synthetic CLI fixtures, three paired repeats with alternating arm order. Binary-capable outer collectors hold baseline printed bytes only inside each trial; only counts and digests are reported. Ordinary loops use scripted decisions and native inference entrypoints are forbidden. The expected new policy is one exact terminal JSON packet and count-only output metadata; interrupt/exit fixtures deliberately produce no packet and are caught outside main to measure restored streams and sanitized integer exit codes. Source exception kind is separate from the propagated exception. Actual model/UI/network calls are zero if the native guard is never hit. This evaluates console transport and exit-code normalization, not semantic quality, interpreter exception printing, an async runtime or a sandbox.",
  "negative_results": "Discard removes console debugging text from the CLI and fixed counters add metadata/context for small-output tasks. Python stream redirection is process-global; saved streams, existing logging handlers, native file descriptors, subprocesses and concurrent/late threads can bypass it and are excluded. Direct SDK console behavior is outside this CLI benchmark. KeyboardInterrupt changes to quiet SystemExit(130); task SystemExit retains exact integers 1..255 but text/zero/other codes become 1. A terminal packet, counter metadata and runtime effects remain unknown for those exits; fixed synthetic effects here are independently checked via local markers and no effects are inferred from exit codes. Outer benchmark catches prevent interpreter printing, so these rows alone do not prove actual interpreter output. Baseline terminal recovery after pollution is diagnostic only and does not count as parseable output. No model accuracy, speed or fee gain is claimed.",
  "cost_conversion": "No tariff or savings assumed. Actual usage is zero only with zero native guard attempts; guard hits produce unknown usage and a failed policy result. Whole-operation timing includes CLI parsing, trusted import, stream capture/discard, callback effects, scripted decisions, private synthetic journal and terminal serialization/stream restoration; excludes worker/import/temp-script setup and post-call analysis. Returned context bytes are actual stdout plus stderr bytes, including synthetic baseline pollution. Text counters are Unicode code points and binary counters are bytes; raw bytes/text are never saved in this report.",
  "summary": {
    "baseline": {
      "runs": 57,
      "ordinary_terminal_runs": 42,
      "exact_json_terminals": 12,
      "propagated_base_exceptions": 15,
      "propagated_exception_markers": 6,
      "correct": 6,
      "status_and_effect_correct": 45,
      "terminal_policy_correct": 6,
      "task_output_metadata_correct": 21,
      "streams_restored": 57,
      "raw_marker_exposures": {
        "stdout": 51,
        "stderr": 30
      },
      "native_inference_attempts": 0,
      "scripted_decisions": 6,
      "median_whole_operation_seconds": 0.0012581670016516,
      "returned_context_bytes": 3016230,
      "scripted_decision_bytes": 10740,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      }
    },
    "candidate": {
      "runs": 57,
      "ordinary_terminal_runs": 42,
      "exact_json_terminals": 42,
      "propagated_base_exceptions": 15,
      "propagated_exception_markers": 0,
      "correct": 57,
      "status_and_effect_correct": 57,
      "terminal_policy_correct": 57,
      "task_output_metadata_correct": 57,
      "streams_restored": 57,
      "raw_marker_exposures": {
        "stdout": 0,
        "stderr": 0
      },
      "native_inference_attempts": 0,
      "scripted_decisions": 6,
      "median_whole_operation_seconds": 0.0012709580041700974,
      "returned_context_bytes": 19906,
      "scripted_decision_bytes": 10739,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      }
    }
  },
  "rows": [
    {
      "case": "quiet",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a5538fd679b5ca79ef4035d2164fd9fea31792514865816b63ffab57d762a5af",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 317,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "bb78ae9678063dc07d814aaf2a63cd67175b4c5df3f8829b5742f3475a87f701",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0011182919988641515,
      "returned_context_bytes": 317,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "quiet",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a5538fd679b5ca79ef4035d2164fd9fea31792514865816b63ffab57d762a5af",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "9a57bfe6b3fbef817f15040dff8ee47cb7a7a99866a46afc892485577935a1f4",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0011977090034633875,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "quiet",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a5538fd679b5ca79ef4035d2164fd9fea31792514865816b63ffab57d762a5af",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 317,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "980cb57eb219f86a113556cc9846f1833cab2b48d3f85332d0bc5b9af475a342",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012508749932749197,
      "returned_context_bytes": 317,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "quiet",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a5538fd679b5ca79ef4035d2164fd9fea31792514865816b63ffab57d762a5af",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "f0b0494e5ad7c2233b59f30d2b3145e3f34dbede8a3a5603bff0945d13136e2c",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012182080099591985,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "quiet",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a5538fd679b5ca79ef4035d2164fd9fea31792514865816b63ffab57d762a5af",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "61145f4ff52bad3d7be3aad34b5745e6bd5f20beff02778a46e63206f275101f",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012491249945014715,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "quiet",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a5538fd679b5ca79ef4035d2164fd9fea31792514865816b63ffab57d762a5af",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 317,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "45c94b746a0d679c2d0be70f1441e1a1646973c098edcd9885bcd7f3160978d6",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012998749880352989,
      "returned_context_bytes": 317,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "import_stdout",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e075e46371b715e37f62859e268713ae61025ee9f2e140f8d4e5d5c675e23227",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "22b89cbd2c9d5348e0211850cbef636761ba85d5f602e33ac894b46261af04ee",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012957079889019951,
      "returned_context_bytes": 340,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "import_stdout",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e075e46371b715e37f62859e268713ae61025ee9f2e140f8d4e5d5c675e23227",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "202b7493bd220277605b8df1febce92416e42618cb5b89c815ba9e1954ad7fe2",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012709580041700974,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "import_stdout",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e075e46371b715e37f62859e268713ae61025ee9f2e140f8d4e5d5c675e23227",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "b8d3413879cc026bfb0588f9620fef7fe379945124a280fcf72eecc211da932a",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012592090060934424,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "import_stdout",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e075e46371b715e37f62859e268713ae61025ee9f2e140f8d4e5d5c675e23227",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "5a0b17e4a6a8406eb41e8d248440aea463eff16a114c6f54fad1d945cf4c27be",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012476249976316467,
      "returned_context_bytes": 340,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "import_stdout",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e075e46371b715e37f62859e268713ae61025ee9f2e140f8d4e5d5c675e23227",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 339,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "91d94895bd01eb07d180791bb4c4be6e5980e6af2fac2a00436f159cb133bdec",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012803749996237457,
      "returned_context_bytes": 339,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "import_stdout",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e075e46371b715e37f62859e268713ae61025ee9f2e140f8d4e5d5c675e23227",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 479,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "3586c3097c096f048fdc45a78713ce66c2c3876704b7df688f4e8fbdeb1b8a3e",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001215875003254041,
      "returned_context_bytes": 479,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "import_stderr",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "af6606de75b28b268585fad8b7b6fd203035d260b4bf66c4a666ff33287f26b3",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 1
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "1a56d914660131bbb13289bd0e61b1fb403d80e4a70100bf9e9fa0b97afb975a",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0013214170030551031,
      "returned_context_bytes": 340,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "import_stderr",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "af6606de75b28b268585fad8b7b6fd203035d260b4bf66c4a666ff33287f26b3",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "cf7f171af255776ac1c6caacd6ba180599c0f54ce675fc714ea723d2154de7c8",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001371416001347825,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "import_stderr",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "af6606de75b28b268585fad8b7b6fd203035d260b4bf66c4a666ff33287f26b3",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "bf97b73ef85a604023e695bae8872634f900938f0a2a8290a17df83df93208a4",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012780830002157018,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "import_stderr",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "af6606de75b28b268585fad8b7b6fd203035d260b4bf66c4a666ff33287f26b3",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 1
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "eb36b2a9168a4ea03aa38c7929d71ff3c1960349ec88c2488e92a147c0a1b29d",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0011695409921230748,
      "returned_context_bytes": 340,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "import_stderr",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "af6606de75b28b268585fad8b7b6fd203035d260b4bf66c4a666ff33287f26b3",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 1
        },
        "captured_stdout_bytes": 317,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "0ced0b95ad2e66d2d2b0d05e9bfd23a6936e879ccb63ca83a8b1b55bcc30e8e9",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0011866669956361875,
      "returned_context_bytes": 339,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "import_stderr",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "af6606de75b28b268585fad8b7b6fd203035d260b4bf66c4a666ff33287f26b3",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "3d469920654e5ca88baaaababb2be4ae78d019d7bebfe5ee0ee7746cc8515a9f",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012933340040035546,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "import_both",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "3af7aefaef00189ef798c4d44831ff859b213e88760ffeb9b1c8b69deb3f9bc4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "4c1ccd6f49010cf60874da62d54c8d0113936aa6b656dce241608e37c442ebbc",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012732920004054904,
      "returned_context_bytes": 362,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "import_both",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "3af7aefaef00189ef798c4d44831ff859b213e88760ffeb9b1c8b69deb3f9bc4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 481,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "f5bd583861834b2ed26cf705c23a353828212d0494ceeaab8a811c367b0d8a78",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012630000128410757,
      "returned_context_bytes": 481,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "import_both",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "3af7aefaef00189ef798c4d44831ff859b213e88760ffeb9b1c8b69deb3f9bc4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "f4f7b932ab92a6d7b1bc322a513ca92e64c5796a05d86395fbcf7d62f68b1387",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001221083992277272,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "import_both",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "3af7aefaef00189ef798c4d44831ff859b213e88760ffeb9b1c8b69deb3f9bc4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 339,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "e32cdb3682617d2345ab921ce9bf1c6aa60248db77ac3f1ed248953791503f90",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012803749996237457,
      "returned_context_bytes": 361,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "import_both",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "3af7aefaef00189ef798c4d44831ff859b213e88760ffeb9b1c8b69deb3f9bc4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "e3bdc0fb8a5121ce864a0fbb1a62e20df405d244717ba989802d3114c471032a",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012512499961303547,
      "returned_context_bytes": 362,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "import_both",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "3af7aefaef00189ef798c4d44831ff859b213e88760ffeb9b1c8b69deb3f9bc4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "7bc66dabfc3bae91b724f58f0b1487b358713a4b2d44d6cc24667800a341e597",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013482080103131011,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "observer_verifier",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "cfa1753b38c510713f8d29f4c1c6b055e66d2d69c4b1356d43d8e2696083bae1",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "82232ae8fff472291327adaadd7691f7fcad8c9c8c83a9ee3063acd2d2af920c",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0013412499974947423,
      "returned_context_bytes": 362,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "observer_verifier",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "cfa1753b38c510713f8d29f4c1c6b055e66d2d69c4b1356d43d8e2696083bae1",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 481,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "8cc827756b00b97942228e6a3b9782e72b2f4d9e1abebb28e6e14627d98c772b",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013213329948484898,
      "returned_context_bytes": 481,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "observer_verifier",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "cfa1753b38c510713f8d29f4c1c6b055e66d2d69c4b1356d43d8e2696083bae1",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 481,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "52ab8c410672dfafbe1818f527f4e8253d5a37ede7ded716e98a3de84c462650",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013419580063782632,
      "returned_context_bytes": 481,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "observer_verifier",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "cfa1753b38c510713f8d29f4c1c6b055e66d2d69c4b1356d43d8e2696083bae1",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "c04666ca9f64db8130c1c7a85b5cf44556c825e4568acd4f68cfd477c9ec4615",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012771250039804727,
      "returned_context_bytes": 362,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "observer_verifier",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "cfa1753b38c510713f8d29f4c1c6b055e66d2d69c4b1356d43d8e2696083bae1",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "e63aff37fc7566b27f1165954f2727a0db3ad2758f61ba82d7175b81737608b2",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012025000032735988,
      "returned_context_bytes": 362,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "observer_verifier",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "cfa1753b38c510713f8d29f4c1c6b055e66d2d69c4b1356d43d8e2696083bae1",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 481,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "ba53cb51d2b8802a6c62edf7f5d4080e330a638dfe4d4f98cddf4382ef7d4adb",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012717909994535148,
      "returned_context_bytes": 481,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "success_effect",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "252880c49a68bd402055ef9b57c9d626c757c68c175eb343995e6704406c5bc8",
      "expected": {
        "status": "done",
        "effects": 1,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 545,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "7d87552ed655230ef965f07d7d70bd0b49cca998790ee92a7d1b44c5748f516c",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.001989500000490807,
      "returned_context_bytes": 567,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "success_effect",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "252880c49a68bd402055ef9b57c9d626c757c68c175eb343995e6704406c5bc8",
      "expected": {
        "status": "done",
        "effects": 1,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 686,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "ec38bc86b013f1f126e902c3dc8dbfd821c84793aa3bd243767fabb694943423",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0018262919911649078,
      "returned_context_bytes": 686,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "success_effect",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "252880c49a68bd402055ef9b57c9d626c757c68c175eb343995e6704406c5bc8",
      "expected": {
        "status": "done",
        "effects": 1,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 686,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "03a3a7efb68ab7d6204dea7da3329b9b90fecce1e630231e1ab531c2a6080d16",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0018302079988643527,
      "returned_context_bytes": 686,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "success_effect",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "252880c49a68bd402055ef9b57c9d626c757c68c175eb343995e6704406c5bc8",
      "expected": {
        "status": "done",
        "effects": 1,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 545,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "588ab931001058485a1e26811cd5b674e1ad67d1250dffc23666ebcf7b94b2c0",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.001801666003302671,
      "returned_context_bytes": 567,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "success_effect",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "252880c49a68bd402055ef9b57c9d626c757c68c175eb343995e6704406c5bc8",
      "expected": {
        "status": "done",
        "effects": 1,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 544,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "afa616c6dcd0e09dadb68d57c0e32cd1460cff82dc9148edbcc9849d749a10f8",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0018312910106033087,
      "returned_context_bytes": 566,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "success_effect",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "252880c49a68bd402055ef9b57c9d626c757c68c175eb343995e6704406c5bc8",
      "expected": {
        "status": "done",
        "effects": 1,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 685,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "2f4aa1c823bc9124aa451e787c3b883776a293a57714df687f6207e10e206c19",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001739457991789095,
      "returned_context_bytes": 685,
      "scripted_decision_bytes": 1789,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "uncertain_effect",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ca1d56d991d96ec813be8ae0efa2312ff2d690d00c330264bcb06575507e7d32",
      "expected": {
        "status": "execution_uncertain",
        "effects": 1,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "execution_uncertain",
        "exit_code": 2,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "TimeoutError",
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 615,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "1793e9bb4679304cc56a033ff7d3f826867ad43e967042c2cb29b2e44505857d",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0017827920091804117,
      "returned_context_bytes": 637,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "uncertain_effect",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ca1d56d991d96ec813be8ae0efa2312ff2d690d00c330264bcb06575507e7d32",
      "expected": {
        "status": "execution_uncertain",
        "effects": 1,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "execution_uncertain",
        "exit_code": 2,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "TimeoutError",
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 755,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "81fde9fb1a17c75cd6606b2f9dcac6678b3fcd1d11a944ac2ff0b15cbe8174c9",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0019244580034865066,
      "returned_context_bytes": 755,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "uncertain_effect",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ca1d56d991d96ec813be8ae0efa2312ff2d690d00c330264bcb06575507e7d32",
      "expected": {
        "status": "execution_uncertain",
        "effects": 1,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "execution_uncertain",
        "exit_code": 2,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "TimeoutError",
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 756,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "a15185a42a0c338d707b333d921487231b2795c721316581b61109e62753043b",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001702499997918494,
      "returned_context_bytes": 756,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "uncertain_effect",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ca1d56d991d96ec813be8ae0efa2312ff2d690d00c330264bcb06575507e7d32",
      "expected": {
        "status": "execution_uncertain",
        "effects": 1,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "execution_uncertain",
        "exit_code": 2,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "TimeoutError",
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 616,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "a7a8d9f7651034db04f196526df8b97b194c18beb3f4f7dc78ddce0fb1c121d6",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0016829999949550256,
      "returned_context_bytes": 638,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "uncertain_effect",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ca1d56d991d96ec813be8ae0efa2312ff2d690d00c330264bcb06575507e7d32",
      "expected": {
        "status": "execution_uncertain",
        "effects": 1,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "execution_uncertain",
        "exit_code": 2,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "TimeoutError",
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 615,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "e0b38349b3ba268aed8268e2169e3c286ccba9a634b14161036f589a9d764af7",
        "captured_stderr_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0017754580039763823,
      "returned_context_bytes": 637,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "uncertain_effect",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ca1d56d991d96ec813be8ae0efa2312ff2d690d00c330264bcb06575507e7d32",
      "expected": {
        "status": "execution_uncertain",
        "effects": 1,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "execution_uncertain",
        "exit_code": 2,
        "effects": 1,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "TimeoutError",
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 22,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 756,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "20e671109de99db2219bb59ce3f885e55475695811626ed8d6986a9c4ce117cf",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 1,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001855958005762659,
      "returned_context_bytes": 756,
      "scripted_decision_bytes": 1790,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "import_exception",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "4897ee83970869f10ef32d4c2a780c9f8c4d98d0eddd23542830c641a659b131",
      "expected": {
        "status": "failed",
        "effects": 0,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "failed",
        "exit_code": 2,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "ValueError",
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 90,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "03b113486e747e7f6082983b9938841140520c537c91ecda731ac05d17e449a4",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0010504170058993623,
      "returned_context_bytes": 90,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "import_exception",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "4897ee83970869f10ef32d4c2a780c9f8c4d98d0eddd23542830c641a659b131",
      "expected": {
        "status": "failed",
        "effects": 0,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "failed",
        "exit_code": 2,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "ValueError",
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 230,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "bf9784abb1197dd5fdb5ab686b5bdf5501b2f04d5bc7d39f652f236cf7ca0c50",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0011249159870203584,
      "returned_context_bytes": 230,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "import_exception",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "4897ee83970869f10ef32d4c2a780c9f8c4d98d0eddd23542830c641a659b131",
      "expected": {
        "status": "failed",
        "effects": 0,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "failed",
        "exit_code": 2,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "ValueError",
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 230,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "bf9784abb1197dd5fdb5ab686b5bdf5501b2f04d5bc7d39f652f236cf7ca0c50",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0009204160014633089,
      "returned_context_bytes": 230,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "import_exception",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "4897ee83970869f10ef32d4c2a780c9f8c4d98d0eddd23542830c641a659b131",
      "expected": {
        "status": "failed",
        "effects": 0,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "failed",
        "exit_code": 2,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "ValueError",
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 90,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "03b113486e747e7f6082983b9938841140520c537c91ecda731ac05d17e449a4",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0013005419896217063,
      "returned_context_bytes": 90,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "import_exception",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "4897ee83970869f10ef32d4c2a780c9f8c4d98d0eddd23542830c641a659b131",
      "expected": {
        "status": "failed",
        "effects": 0,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "failed",
        "exit_code": 2,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "ValueError",
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 90,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "03b113486e747e7f6082983b9938841140520c537c91ecda731ac05d17e449a4",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0008694999996805564,
      "returned_context_bytes": 90,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "import_exception",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "4897ee83970869f10ef32d4c2a780c9f8c4d98d0eddd23542830c641a659b131",
      "expected": {
        "status": "failed",
        "effects": 0,
        "exit_code": 2,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "failed",
        "exit_code": 2,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": "ValueError",
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 22,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 230,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "bf9784abb1197dd5fdb5ab686b5bdf5501b2f04d5bc7d39f652f236cf7ca0c50",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0009074580011656508,
      "returned_context_bytes": 230,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "unicode",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "32887dc20dbc868eff51c1f3ed1d3e63210a237711dd743e7daf98dd3f8661f0",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 349,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "ea3d76eb2cc0dddc876b8301a871c498e350b4a2c775517e12c420b2818ea244",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012603329960256815,
      "returned_context_bytes": 349,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "unicode",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "32887dc20dbc868eff51c1f3ed1d3e63210a237711dd743e7daf98dd3f8661f0",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "99064af582208a3710153eaf80f1da254bdc996ce72469080d55a60cb0bf3095",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012365000002318993,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "unicode",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "32887dc20dbc868eff51c1f3ed1d3e63210a237711dd743e7daf98dd3f8661f0",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 479,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "648bf11ae9821bdb6d6bfcca901dad5f5e87240b31db2909832a78a5a59dd481",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013118330098222941,
      "returned_context_bytes": 479,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "unicode",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "32887dc20dbc868eff51c1f3ed1d3e63210a237711dd743e7daf98dd3f8661f0",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 347,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "015ac51767c15fa90da4ca65b94c0e33704a8f50bc8bbd6cc4cb713d4c19e329",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0013277499965624884,
      "returned_context_bytes": 347,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "unicode",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "32887dc20dbc868eff51c1f3ed1d3e63210a237711dd743e7daf98dd3f8661f0",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 349,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "694896a109ed248615db2b80c7a7eb11ed3f1434877c813a9e2785889bf6c862",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.001401375004206784,
      "returned_context_bytes": 349,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "unicode",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "32887dc20dbc868eff51c1f3ed1d3e63210a237711dd743e7daf98dd3f8661f0",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 24,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "b111e77be77fbda6cf7a487872cb1da21445baea4889864a4f4abadb4c93877d",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013339580036699772,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "binary_invalid_utf8",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "65eba13f2393d5219c8937cd4a4099fa22fefc38c1f2157dc7c15ce9e37ac2e4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "50e8071c2d086ccc08716d310a48c1c2bfe16751754419e61ef4dcb32eb1cdc6",
        "captured_stderr_sha256": "e5be02cbf4b9509ee0c28135fac8bafba06fda48c35090898f1f44104ea7ac11",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0013362500030780211,
      "returned_context_bytes": 362,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "binary_invalid_utf8",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "65eba13f2393d5219c8937cd4a4099fa22fefc38c1f2157dc7c15ce9e37ac2e4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "2df4886924f99bb9b59bcf4bfc138322b9ec217e8ba6c650ccf8261020229285",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012664999958360568,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "binary_invalid_utf8",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "65eba13f2393d5219c8937cd4a4099fa22fefc38c1f2157dc7c15ce9e37ac2e4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 480,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "10539df9919532d87776db49a47f666a093ff30f506134ad9bb3679f40fa2b32",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012283329997444525,
      "returned_context_bytes": 480,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "binary_invalid_utf8",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "65eba13f2393d5219c8937cd4a4099fa22fefc38c1f2157dc7c15ce9e37ac2e4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "e1ac948299ce867a3e0151872ef5d2950a80654833c66d09cb27ce257908f8ca",
        "captured_stderr_sha256": "e5be02cbf4b9509ee0c28135fac8bafba06fda48c35090898f1f44104ea7ac11",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012129170063417405,
      "returned_context_bytes": 362,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "binary_invalid_utf8",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "65eba13f2393d5219c8937cd4a4099fa22fefc38c1f2157dc7c15ce9e37ac2e4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 1
        },
        "captured_stdout_bytes": 340,
        "captured_stderr_bytes": 22,
        "captured_stdout_sha256": "f53df0fc958200217b5d315e094a8060edde224c155852f32d55f0380d6b6e16",
        "captured_stderr_sha256": "e5be02cbf4b9509ee0c28135fac8bafba06fda48c35090898f1f44104ea7ac11",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0013272080104798079,
      "returned_context_bytes": 362,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "binary_invalid_utf8",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "65eba13f2393d5219c8937cd4a4099fa22fefc38c1f2157dc7c15ce9e37ac2e4",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 481,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "f65d97f6d2686d57bd192901a8af814c1030e3459da73307e96ac208b178c39d",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012454579991754144,
      "returned_context_bytes": 481,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "large_output",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "471e9d206b2b6b2340d8c06d16f660d64b6924323430bbe2b64bc645dde37951",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 1000339,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "9375e63040dce6835e1d4645c844d29c748cb2d683ead2cbcad3e3344de478c7",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0016321670118486509,
      "returned_context_bytes": 1000339,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "large_output",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "471e9d206b2b6b2340d8c06d16f660d64b6924323430bbe2b64bc645dde37951",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 484,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "5b11e06bb1eb73722ca6f202feb62224c56c551e10cd8f75830660d178de50c6",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001312583000981249,
      "returned_context_bytes": 484,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "large_output",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "471e9d206b2b6b2340d8c06d16f660d64b6924323430bbe2b64bc645dde37951",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 483,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "5e5b439366d04ab9834579f0d3ea56917b169604b9be884a3e4b6b47ad2b1325",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013185840070946142,
      "returned_context_bytes": 483,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "large_output",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "471e9d206b2b6b2340d8c06d16f660d64b6924323430bbe2b64bc645dde37951",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 1000340,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "8f0e28f9a65491e779e981e7736903089b0a89a0f8d419d299b29da6aec08ba1",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0016985840047709644,
      "returned_context_bytes": 1000340,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "large_output",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "471e9d206b2b6b2340d8c06d16f660d64b6924323430bbe2b64bc645dde37951",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 1000339,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "7a497812db807f1b899806c66b4c9c2dde2436fe5a2f0abc2e693455f772427e",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.001658124994719401,
      "returned_context_bytes": 1000339,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "large_output",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "471e9d206b2b6b2340d8c06d16f660d64b6924323430bbe2b64bc645dde37951",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 1000022,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 0,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 485,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "22d78d10da92339451b5e60cfabdba05ece216684855bb4f5042599236e087b4",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0014119580009719357,
      "returned_context_bytes": 485,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "empty_writes",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ddfad6c0f12d9f423b7b23441b77164e65c555dd297d7b939737295eac571ea7",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "36bfd117b2a8187c5cb8c0e352b1335cee89a51be9ab81b578ec94b19beb632b",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012405000015860423,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "empty_writes",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ddfad6c0f12d9f423b7b23441b77164e65c555dd297d7b939737295eac571ea7",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "d0654f181b328d61c181d9b04b7664a02d6144120e830ddb6cb78c7856456514",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013860829931218177,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "empty_writes",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ddfad6c0f12d9f423b7b23441b77164e65c555dd297d7b939737295eac571ea7",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "8c7fce2643897e964b0d21945691e45691d352df91dd59e62fe1b123197ad86b",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013558330101659521,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "empty_writes",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ddfad6c0f12d9f423b7b23441b77164e65c555dd297d7b939737295eac571ea7",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "2560d4b37fa9efb2e29c7b53c36571e558610e183eed768d9ff5378ea0288342",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013231250050012022,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "empty_writes",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ddfad6c0f12d9f423b7b23441b77164e65c555dd297d7b939737295eac571ea7",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "d80fb0fe3bb33ee0dfae2299e72b5ab03a706e8f2c2f55d8d6e176350c53fde9",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0012208750122226775,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "empty_writes",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "ddfad6c0f12d9f423b7b23441b77164e65c555dd297d7b939737295eac571ea7",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "fa9dd3a639c64c077ab11a1c81a09eb6f6f1e95ba17e3ec8b6d889d3f842d262",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001299290990573354,
      "returned_context_bytes": 318,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "warning",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "d87f7444991978e67a9d067c6a6be5f299ce2b7c3ebef464eb3baf6a6edd85fe",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 2
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 168,
        "captured_stdout_sha256": "1603b082b4d705574f918920e3b1c1801794bc6adda471bef2b41d7937914733",
        "captured_stderr_sha256": "a0ac664729560aa2e07da9d822d79fdc5af9f411f76dd6e183e9ebe1128ed95a",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012581670016516,
      "returned_context_bytes": 486,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "warning",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "d87f7444991978e67a9d067c6a6be5f299ce2b7c3ebef464eb3baf6a6edd85fe",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 481,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "7dc7a3bde5d2ed540367766e455cd133fc399ce3d9725561bd95a1e413b2268b",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013822500040987507,
      "returned_context_bytes": 481,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "warning",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "d87f7444991978e67a9d067c6a6be5f299ce2b7c3ebef464eb3baf6a6edd85fe",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 481,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "aabc8c1a82977b34c18f67cd91daee9df634587ec6dfefd2c9ec2b9785c522bb",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013688329927390441,
      "returned_context_bytes": 481,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "warning",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "d87f7444991978e67a9d067c6a6be5f299ce2b7c3ebef464eb3baf6a6edd85fe",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 2
        },
        "captured_stdout_bytes": 317,
        "captured_stderr_bytes": 168,
        "captured_stdout_sha256": "79c816311563b7c6ba8b5fc16540f647b192c3064267854dd9539a7c190344ad",
        "captured_stderr_sha256": "d5478d3f5c8879cacf1d631dd8260a0e93040daf0358608fe2c617a48789fa71",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.001316042005782947,
      "returned_context_bytes": 485,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "warning",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "d87f7444991978e67a9d067c6a6be5f299ce2b7c3ebef464eb3baf6a6edd85fe",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 2
        },
        "captured_stdout_bytes": 318,
        "captured_stderr_bytes": 168,
        "captured_stdout_sha256": "b74eb33a043bd4673b5dcae57a4a83f5cfcbe4e4aa84f93060471833b85fbce7",
        "captured_stderr_sha256": "66fb00f41632c62cffba11177afa4a36b036197fcc25f01acde4265e50323e9e",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0013204160059103742,
      "returned_context_bytes": 486,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "warning",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "d87f7444991978e67a9d067c6a6be5f299ce2b7c3ebef464eb3baf6a6edd85fe",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 0,
            "binary_bytes": 0
          },
          "stderr": {
            "characters": 168,
            "binary_bytes": 0
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 481,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "fbc2e0b79e1546d626d1d0b4ee753c4ab5f112443e646da887656a12990acfa0",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001258208998478949,
      "returned_context_bytes": 481,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "write_return_counts",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "94a1378596aef85aba6c09321c40642cc8865354cd8c2de3a5f6e982ab243e0e",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 3,
          "stderr": 2
        },
        "captured_stdout_bytes": 389,
        "captured_stderr_bytes": 43,
        "captured_stdout_sha256": "eda97d6733ba19809ad162038701fdc914172ee6e912be92893b040af2c87368",
        "captured_stderr_sha256": "d6a6a23c4caac9c69f8c898acba4341a5420731199462ecf70458a391cc0d3b3",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0012767089938279241,
      "returned_context_bytes": 432,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "write_return_counts",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "94a1378596aef85aba6c09321c40642cc8865354cd8c2de3a5f6e982ab243e0e",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 483,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "1f028c49784b2a17376be37fffc21789ba7793277288f8e2279565bdff510ef6",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.001356790991849266,
      "returned_context_bytes": 483,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "write_return_counts",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "94a1378596aef85aba6c09321c40642cc8865354cd8c2de3a5f6e982ab243e0e",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 482,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "b2f4b1f5ba18b90dd3a00b526e26d452b075af919cc2648012bd81bdc52e3ed1",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013247079914435744,
      "returned_context_bytes": 482,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "write_return_counts",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "94a1378596aef85aba6c09321c40642cc8865354cd8c2de3a5f6e982ab243e0e",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 3,
          "stderr": 2
        },
        "captured_stdout_bytes": 389,
        "captured_stderr_bytes": 43,
        "captured_stdout_sha256": "a042e98d29c43cc47d70e1391e1acc697622a6c11237b1f18070434166c5d814",
        "captured_stderr_sha256": "d6a6a23c4caac9c69f8c898acba4341a5420731199462ecf70458a391cc0d3b3",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.001271083005121909,
      "returned_context_bytes": 432,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "write_return_counts",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "94a1378596aef85aba6c09321c40642cc8865354cd8c2de3a5f6e982ab243e0e",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": false,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 3,
          "stderr": 2
        },
        "captured_stdout_bytes": 389,
        "captured_stderr_bytes": 43,
        "captured_stdout_sha256": "d01e9558232b18994d3072fcf1758fe93a6201ac85ec66a524e8af305deedea7",
        "captured_stderr_sha256": "d6a6a23c4caac9c69f8c898acba4341a5420731199462ecf70458a391cc0d3b3",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0013009579997742549,
      "returned_context_bytes": 432,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "write_return_counts",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "94a1378596aef85aba6c09321c40642cc8865354cd8c2de3a5f6e982ab243e0e",
      "expected": {
        "status": "done",
        "effects": 0,
        "exit_code": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null
      },
      "result": {
        "status": "done",
        "exit_code": 0,
        "effects": 0,
        "base_exception": null,
        "source_exception_kind": null,
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": true,
        "terminal_packet_recovered": true,
        "terminal_policy_correct": true,
        "task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "expected_task_output_metadata": {
          "policy": "discard",
          "scope": "python_streams",
          "stdout": {
            "characters": 45,
            "binary_bytes": 22
          },
          "stderr": {
            "characters": 21,
            "binary_bytes": 22
          }
        },
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 483,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "c8c94af8eefe82203f456875535e22ff8a4e6b5513a55d5cfa4645c767d63e53",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0013020420010434464,
      "returned_context_bytes": 483,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "system_exit",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "67310d7bc6fc13ddd00e092930d248cab3a598c9f8c39d8fa2e31de3a4999dc1",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0008950000046752393,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "system_exit",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "67310d7bc6fc13ddd00e092930d248cab3a598c9f8c39d8fa2e31de3a4999dc1",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0009630830027163029,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "system_exit",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "67310d7bc6fc13ddd00e092930d248cab3a598c9f8c39d8fa2e31de3a4999dc1",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0009202079963870347,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "system_exit",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "67310d7bc6fc13ddd00e092930d248cab3a598c9f8c39d8fa2e31de3a4999dc1",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0009268750000046566,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "system_exit",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "67310d7bc6fc13ddd00e092930d248cab3a598c9f8c39d8fa2e31de3a4999dc1",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0009058339928742498,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "system_exit",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "67310d7bc6fc13ddd00e092930d248cab3a598c9f8c39d8fa2e31de3a4999dc1",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 7,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0009105000062845647,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "keyboard_interrupt",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "53bf3a022bfbfbdaae451c4020238cc3cd26ac9ebe7ad51ea93216f6a27e2cd6",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "KeyboardInterrupt",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0009233330056304112,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "keyboard_interrupt",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "53bf3a022bfbfbdaae451c4020238cc3cd26ac9ebe7ad51ea93216f6a27e2cd6",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0009125420037889853,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "keyboard_interrupt",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "53bf3a022bfbfbdaae451c4020238cc3cd26ac9ebe7ad51ea93216f6a27e2cd6",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0008784170058788732,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "keyboard_interrupt",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "53bf3a022bfbfbdaae451c4020238cc3cd26ac9ebe7ad51ea93216f6a27e2cd6",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "KeyboardInterrupt",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.000905084001715295,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "keyboard_interrupt",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "53bf3a022bfbfbdaae451c4020238cc3cd26ac9ebe7ad51ea93216f6a27e2cd6",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "KeyboardInterrupt",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0009188749972963706,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "keyboard_interrupt",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "53bf3a022bfbfbdaae451c4020238cc3cd26ac9ebe7ad51ea93216f6a27e2cd6",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.002058583006146364,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "system_exit_text",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a316c137b38ecdb618ed7f043c59ab2c65a6acc636cd55b23e57713e61a0954b",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": null,
        "exception_code_type": "str",
        "propagated_exception_marker_count": 1,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0011477499938337132,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "system_exit_text",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a316c137b38ecdb618ed7f043c59ab2c65a6acc636cd55b23e57713e61a0954b",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0009321249963250011,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "system_exit_text",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a316c137b38ecdb618ed7f043c59ab2c65a6acc636cd55b23e57713e61a0954b",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.00105779200384859,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "system_exit_text",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a316c137b38ecdb618ed7f043c59ab2c65a6acc636cd55b23e57713e61a0954b",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": null,
        "exception_code_type": "str",
        "propagated_exception_marker_count": 1,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0010350829979870468,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "system_exit_text",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a316c137b38ecdb618ed7f043c59ab2c65a6acc636cd55b23e57713e61a0954b",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": null,
        "exception_code_type": "str",
        "propagated_exception_marker_count": 1,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0008970830094767734,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "system_exit_text",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "a316c137b38ecdb618ed7f043c59ab2c65a6acc636cd55b23e57713e61a0954b",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0008929579926189035,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "system_exit_zero",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "824bf0762cdb869039cb8b73811f3d8810c86b7c5d276bf5ce208e83fbbdfb55",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 0,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0008851670136209577,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "system_exit_zero",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "824bf0762cdb869039cb8b73811f3d8810c86b7c5d276bf5ce208e83fbbdfb55",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0008952500065788627,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "system_exit_zero",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "824bf0762cdb869039cb8b73811f3d8810c86b7c5d276bf5ce208e83fbbdfb55",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0011443750117905438,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "system_exit_zero",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "824bf0762cdb869039cb8b73811f3d8810c86b7c5d276bf5ce208e83fbbdfb55",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 0,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.000861459004227072,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "system_exit_zero",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "824bf0762cdb869039cb8b73811f3d8810c86b7c5d276bf5ce208e83fbbdfb55",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 0,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0010584999981801957,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "system_exit_zero",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "824bf0762cdb869039cb8b73811f3d8810c86b7c5d276bf5ce208e83fbbdfb55",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "SystemExit",
        "exception_exit_code": 1,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0009432499937247485,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    },
    {
      "case": "keyboard_interrupt_text",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e9bfb375c06dda959105ff620c104b566672c1c0a5ca314dd27d6189af2dea96",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "KeyboardInterrupt",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 1,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0008872920006979257,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 1
    },
    {
      "case": "keyboard_interrupt_text",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e9bfb375c06dda959105ff620c104b566672c1c0a5ca314dd27d6189af2dea96",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.000877583006513305,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 1
    },
    {
      "case": "keyboard_interrupt_text",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e9bfb375c06dda959105ff620c104b566672c1c0a5ca314dd27d6189af2dea96",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0008979590056696907,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 2
    },
    {
      "case": "keyboard_interrupt_text",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e9bfb375c06dda959105ff620c104b566672c1c0a5ca314dd27d6189af2dea96",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "KeyboardInterrupt",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 1,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0009062080061994493,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 2
    },
    {
      "case": "keyboard_interrupt_text",
      "runtime_version": "0.1.6",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e9bfb375c06dda959105ff620c104b566672c1c0a5ca314dd27d6189af2dea96",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "KeyboardInterrupt",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": null,
        "exception_code_type": null,
        "propagated_exception_marker_count": 1,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": false,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": false,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 1,
          "stderr": 0
        },
        "captured_stdout_bytes": 22,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "dedc44b5988cfa2c224361ac9c405d97ae737aee3d342dea6e0e1fb25385a255",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": false,
      "whole_operation_seconds": 0.0009859169949777424,
      "returned_context_bytes": 22,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "baseline",
      "repeat": 3
    },
    {
      "case": "keyboard_interrupt_text",
      "runtime_version": "0.1.7",
      "runtime_loaded_from_requested_source": true,
      "fixture_sha256": "e9bfb375c06dda959105ff620c104b566672c1c0a5ca314dd27d6189af2dea96",
      "expected": {
        "status": "base_exception",
        "effects": 0,
        "exit_code": null,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130
      },
      "result": {
        "status": "base_exception",
        "exit_code": null,
        "effects": 0,
        "base_exception": "SystemExit",
        "source_exception_kind": "KeyboardInterrupt",
        "exception_exit_code": 130,
        "exception_code_type": "int",
        "propagated_exception_marker_count": 0,
        "error_type": null,
        "terminal_json_exact": false,
        "terminal_packet_recovered": false,
        "terminal_policy_correct": true,
        "task_output_metadata": null,
        "expected_task_output_metadata": null,
        "task_output_metadata_correct": true,
        "status_and_effect_correct": true,
        "streams_restored": true,
        "raw_marker_exposures": {
          "stdout": 0,
          "stderr": 0
        },
        "captured_stdout_bytes": 0,
        "captured_stderr_bytes": 0,
        "captured_stdout_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "captured_stderr_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        "scripted_decisions": 0,
        "native_inference_attempts": 0
      },
      "correct": true,
      "whole_operation_seconds": 0.0008786660037003458,
      "returned_context_bytes": 0,
      "scripted_decision_bytes": 0,
      "actual_model_usage": {
        "requests": 0,
        "models": {},
        "input_tokens": 0,
        "output_tokens": 0,
        "usage_complete": true
      },
      "arm": "candidate",
      "repeat": 3
    }
  ]
}