{
  "schema": "rlebench/acceptance-delivery/1",
  "artifact_acceptance_passed": true,
  "errors": [],
  "episodes": [
    {
      "harness": "codex",
      "episode": 1,
      "seed": 0,
      "native_success": false,
      "native_evidence_valid": true,
      "execution_exception": null,
      "termination": "stopped",
      "steps": 2386,
      "states_verified": 2388,
      "learned_files": [
        "memos/robocasa.md",
        "tools/panda.py"
      ],
      "workspace_files_verified": 4,
      "provider_responses": null
    },
    {
      "harness": "codex",
      "episode": 2,
      "seed": 1,
      "native_success": true,
      "native_evidence_valid": true,
      "execution_exception": null,
      "termination": "success",
      "steps": 794,
      "states_verified": 796,
      "learned_files": [
        "memos/robocasa.md",
        "tools/panda.py"
      ],
      "workspace_files_verified": 4,
      "provider_responses": null
    },
    {
      "harness": "codex",
      "episode": 3,
      "seed": 2,
      "native_success": false,
      "native_evidence_valid": true,
      "execution_exception": null,
      "termination": "stopped",
      "steps": 4621,
      "states_verified": 4623,
      "learned_files": [
        "memos/robocasa.md",
        "tools/panda.py",
        "tools/pivot.py"
      ],
      "workspace_files_verified": 5,
      "provider_responses": null
    },
    {
      "harness": "codex",
      "episode": 4,
      "seed": 3,
      "native_success": true,
      "native_evidence_valid": true,
      "execution_exception": null,
      "termination": "success",
      "steps": 989,
      "states_verified": 991,
      "learned_files": [
        "memos/robocasa.md",
        "tools/panda.py",
        "tools/pivot.py"
      ],
      "workspace_files_verified": 5,
      "provider_responses": null
    },
    {
      "harness": "codex",
      "episode": 5,
      "seed": 4,
      "native_success": true,
      "native_evidence_valid": true,
      "execution_exception": null,
      "termination": "success",
      "steps": 4888,
      "states_verified": 4890,
      "learned_files": [
        "memos/robocasa.md",
        "tools/panda.py",
        "tools/pivot.py"
      ],
      "workspace_files_verified": 5,
      "provider_responses": null
    },
    {
      "harness": "kinex",
      "episode": 1,
      "seed": 0,
      "native_success": false,
      "native_evidence_valid": true,
      "execution_exception": {
        "exception_type": "RuntimeError",
        "exception_message": "Kinex exited 1; inspect kinex.log",
        "exception_traceback": "Traceback (most recent call last):\n  File \"/python-environment/lib/python3.12/site-packages/harbor/trial/multi_step.py\", line 164, in _run_step_agent\n    await self._run_agent_phase(\n  File \"/python-environment/lib/python3.12/site-packages/harbor/trial/trial.py\", line 545, in _run_agent_phase\n    await asyncio.wait_for(\n  File \"/usr/lib/python3.12/asyncio/tasks.py\", line 520, in wait_for\n    return await fut\n           ^^^^^^^^^\n  File \"/source/RLE-Bench-inhouse/src/rlebench2/agents/kinex.py\", line 141, in run\n    raise RuntimeError(f\"Kinex exited {result.return_code}; inspect kinex.log\")\nRuntimeError: Kinex exited 1; inspect kinex.log\n",
        "occurred_at": "2026-09-18T17:07:00.942875Z"
      },
      "termination": "100 model-turn limit",
      "steps": 1336,
      "states_verified": 1338,
      "learned_files": [
        ".kinex/tool/base-step/tool.py",
        ".kinex/tool/base-step/tool.yaml",
        ".kinex/tool/hold-steps/tool.py",
        ".kinex/tool/hold-steps/tool.yaml"
      ],
      "workspace_files_verified": 29,
      "provider_responses": 100
    },
    {
      "harness": "kinex",
      "episode": 2,
      "seed": 1,
      "native_success": true,
      "native_evidence_valid": true,
      "execution_exception": null,
      "termination": "success",
      "steps": 1949,
      "states_verified": 1951,
      "learned_files": [
        ".kinex/memos/open-fridge.md",
        ".kinex/tool/base-step/tool.py",
        ".kinex/tool/base-step/tool.yaml",
        ".kinex/tool/hold-steps/tool.py",
        ".kinex/tool/hold-steps/tool.yaml",
        ".kinex/tool/tcp-move/tool.py",
        ".kinex/tool/tcp-move/tool.yaml"
      ],
      "workspace_files_verified": 32,
      "provider_responses": 76
    },
    {
      "harness": "kinex",
      "episode": 3,
      "seed": 2,
      "native_success": false,
      "native_evidence_valid": true,
      "execution_exception": {
        "exception_type": "RuntimeError",
        "exception_message": "Kinex exited 1; inspect kinex.log",
        "exception_traceback": "Traceback (most recent call last):\n  File \"/python-environment/lib/python3.12/site-packages/harbor/trial/multi_step.py\", line 164, in _run_step_agent\n    await self._run_agent_phase(\n  File \"/python-environment/lib/python3.12/site-packages/harbor/trial/trial.py\", line 545, in _run_agent_phase\n    await asyncio.wait_for(\n  File \"/usr/lib/python3.12/asyncio/tasks.py\", line 520, in wait_for\n    return await fut\n           ^^^^^^^^^\n  File \"/source/RLE-Bench-inhouse/src/rlebench2/agents/kinex.py\", line 141, in run\n    raise RuntimeError(f\"Kinex exited {result.return_code}; inspect kinex.log\")\nRuntimeError: Kinex exited 1; inspect kinex.log\n",
        "occurred_at": "2026-09-18T17:37:46.139757Z"
      },
      "termination": "100 model-turn limit",
      "steps": 3763,
      "states_verified": 3765,
      "learned_files": [
        ".kinex/memos/open-fridge.md",
        ".kinex/tool/base-step/tool.py",
        ".kinex/tool/base-step/tool.yaml",
        ".kinex/tool/hold-steps/tool.py",
        ".kinex/tool/hold-steps/tool.yaml",
        ".kinex/tool/tcp-move/tool.py",
        ".kinex/tool/tcp-move/tool.yaml"
      ],
      "workspace_files_verified": 32,
      "provider_responses": 100
    },
    {
      "harness": "kinex",
      "episode": 4,
      "seed": 3,
      "native_success": false,
      "native_evidence_valid": true,
      "execution_exception": {
        "exception_type": "RuntimeError",
        "exception_message": "Kinex exited 1; inspect kinex.log",
        "exception_traceback": "Traceback (most recent call last):\n  File \"/python-environment/lib/python3.12/site-packages/harbor/trial/multi_step.py\", line 164, in _run_step_agent\n    await self._run_agent_phase(\n  File \"/python-environment/lib/python3.12/site-packages/harbor/trial/trial.py\", line 545, in _run_agent_phase\n    await asyncio.wait_for(\n  File \"/usr/lib/python3.12/asyncio/tasks.py\", line 520, in wait_for\n    return await fut\n           ^^^^^^^^^\n  File \"/source/RLE-Bench-inhouse/src/rlebench2/agents/kinex.py\", line 141, in run\n    raise RuntimeError(f\"Kinex exited {result.return_code}; inspect kinex.log\")\nRuntimeError: Kinex exited 1; inspect kinex.log\n",
        "occurred_at": "2026-09-18T17:56:05.299249Z"
      },
      "termination": "100 model-turn limit",
      "steps": 2226,
      "states_verified": 2228,
      "learned_files": [
        ".kinex/memos/open-fridge.md",
        ".kinex/tool/base-step/tool.py",
        ".kinex/tool/base-step/tool.yaml",
        ".kinex/tool/hold-steps/tool.py",
        ".kinex/tool/hold-steps/tool.yaml",
        ".kinex/tool/tcp-move/tool.py",
        ".kinex/tool/tcp-move/tool.yaml"
      ],
      "workspace_files_verified": 32,
      "provider_responses": 100
    },
    {
      "harness": "kinex",
      "episode": 5,
      "seed": 4,
      "native_success": false,
      "native_evidence_valid": true,
      "execution_exception": {
        "exception_type": "RuntimeError",
        "exception_message": "Kinex exited 1; inspect kinex.log",
        "exception_traceback": "Traceback (most recent call last):\n  File \"/python-environment/lib/python3.12/site-packages/harbor/trial/multi_step.py\", line 164, in _run_step_agent\n    await self._run_agent_phase(\n  File \"/python-environment/lib/python3.12/site-packages/harbor/trial/trial.py\", line 545, in _run_agent_phase\n    await asyncio.wait_for(\n  File \"/usr/lib/python3.12/asyncio/tasks.py\", line 520, in wait_for\n    return await fut\n           ^^^^^^^^^\n  File \"/source/RLE-Bench-inhouse/src/rlebench2/agents/kinex.py\", line 141, in run\n    raise RuntimeError(f\"Kinex exited {result.return_code}; inspect kinex.log\")\nRuntimeError: Kinex exited 1; inspect kinex.log\n",
        "occurred_at": "2026-09-18T18:12:31.565392Z"
      },
      "termination": "100 model-turn limit",
      "steps": 3023,
      "states_verified": 3025,
      "learned_files": [
        ".kinex/memos/open-fridge.md",
        ".kinex/tool/base-step/tool.py",
        ".kinex/tool/base-step/tool.yaml",
        ".kinex/tool/hold-steps/tool.py",
        ".kinex/tool/hold-steps/tool.yaml",
        ".kinex/tool/tcp-move/tool.py",
        ".kinex/tool/tcp-move/tool.yaml"
      ],
      "workspace_files_verified": 32,
      "provider_responses": 100
    }
  ],
  "paired_initial_states": [
    {
      "episode": 1,
      "same_model": true,
      "same_scene_metadata_raw": true,
      "same_scene": true,
      "same_initial_state": true,
      "max_initial_state_difference": 0.0,
      "metadata_comparison": "Only groups_containing_sampled_obj membership labels are sorted for comparison; original manifests remain unchanged."
    },
    {
      "episode": 2,
      "same_model": true,
      "same_scene_metadata_raw": false,
      "same_scene": true,
      "same_initial_state": true,
      "max_initial_state_difference": 0.0,
      "metadata_comparison": "Only groups_containing_sampled_obj membership labels are sorted for comparison; original manifests remain unchanged."
    },
    {
      "episode": 3,
      "same_model": true,
      "same_scene_metadata_raw": true,
      "same_scene": true,
      "same_initial_state": true,
      "max_initial_state_difference": 0.0,
      "metadata_comparison": "Only groups_containing_sampled_obj membership labels are sorted for comparison; original manifests remain unchanged."
    },
    {
      "episode": 4,
      "same_model": true,
      "same_scene_metadata_raw": false,
      "same_scene": true,
      "same_initial_state": true,
      "max_initial_state_difference": 0.0,
      "metadata_comparison": "Only groups_containing_sampled_obj membership labels are sorted for comparison; original manifests remain unchanged."
    },
    {
      "episode": 5,
      "same_model": true,
      "same_scene_metadata_raw": true,
      "same_scene": true,
      "same_initial_state": true,
      "max_initial_state_difference": 0.0,
      "metadata_comparison": "Only groups_containing_sampled_obj membership labels are sorted for comparison; original manifests remain unchanged."
    }
  ],
  "session_ids": {
    "kinex": [
      "ses_00mu777iz4000053",
      "ses_00mu77pf440000ir",
      "ses_00mu785qz4000090",
      "ses_00mu78t0hr00003x",
      "ses_00mu79gjia0000xw"
    ],
    "codex": [
      "01a0b566-719a-7ae0-b9c0-520d3a21e312",
      "01a0b56e-c3f6-70c1-9783-9f834a787f98",
      "01a0b572-ddb7-7090-918c-1213c2ece4af",
      "01a0b580-de2a-7863-a517-7d08c0dbaa1f",
      "01a0b585-a9d6-73e3-aa72-9539a7d6f754"
    ]
  },
  "execution_valid": false,
  "source_provenance": {
    "codex": "Documented recovery copy; see report-recovery.json. Binary archives were unchanged.",
    "kinex": "Unmodified formal trial copy."
  },
  "scope": "Artifact coverage is independent of task success and harness exceptions. Full state playback is not command resimulation."
}