{
  "schema": 1,
  "run_id": "essai-03-2026-11-14",
  "suite": "BPB2 v0.3-pilot",
  "model": "gpt-6-astra",
  "reasoning": "ultra",
  "recorded_at": "2026-11-14T19:03:21.101013+00:00",
  "status": "completed-development-run",
  "tracks": {
    "self": {
      "initial": {
        "pm_out_of_100": 97.5,
        "coverage": 20,
        "ready_count": 18,
        "browser_mean_out_of_4": 2.9,
        "cases_with_open_material_failures": [],
        "strong_result_count": 18
      },
      "current": {
        "pm_out_of_100": 97.5,
        "coverage": 20,
        "ready_count": 18,
        "browser_mean_out_of_4": 2.9,
        "cases_with_open_material_failures": [],
        "strong_result_count": 18
      }
    },
    "checked": {
      "initial": {
        "pm_out_of_100": 98.8125,
        "coverage": 20,
        "ready_count": 19,
        "browser_mean_out_of_4": 3.9,
        "cases_with_open_material_failures": [
          "BPB2-11"
        ],
        "strong_result_count": 19
      },
      "current": {
        "pm_out_of_100": 99.875,
        "coverage": 20,
        "ready_count": 20,
        "browser_mean_out_of_4": 4.0,
        "cases_with_open_material_failures": [],
        "strong_result_count": 20
      }
    }
  },
  "first_attempt_timing": {
    "first_start": "2026-11-14T17:35:52Z",
    "last_finish": "2026-11-14T18:51:26Z",
    "elapsed_wall_seconds": 4534.0,
    "sum_of_case_wall_seconds": 11224.726035000002,
    "median_case_wall_seconds": 541.5,
    "minimum_case_wall_seconds": 414.0,
    "maximum_case_wall_seconds": 871.0,
    "scope": "Actual case receipt intervals, with cases run concurrently; not model active time or human effort."
  },
  "assisted_repairs": [
    {
      "case_id": "BPB2-11",
      "cycle": 1,
      "started_at": "2026-11-14T18:47:42Z",
      "finished_at": "2026-11-14T18:50:15.220460Z",
      "wall_seconds": 153.22046,
      "current_raw_sha256": "8329b9f75487de996fc84528d3169b85e85da2f5ab0b26f1cae1d945c25fb248",
      "served_current_sha256": "8f38bfcafae4d67be4086e46f89e205c46e80ac0777ca0feb1ad7c7290ee6ba9",
      "change": "Correct expected-funding summary propagation when scenario assumptions change. Confirmed defect feedback supplied by the separate review; original first artifact and score retained."
    }
  ],
  "human_practitioner_review": false,
  "claude_review": false,
  "measured_human_checking_minutes": null,
  "measured_workplace_savings": null,
  "tokens": null,
  "billed_cost": null,
  "adaptation_stage": "Not run; five later-evidence exercises remain separate from initial /100.",
  "interpretation": "Separate AI grading in the same model family, supported by independent checks and coordinator browser journeys. One development attempt per case; ordinal ratings, no human calibration or matched time-saving comparison. Earlier essais use a different suite."
}
