{
  "benchmark_version": "2.3.0",
  "study_title": "Thunderbear Agentic Mastery v2.3 & Resonant Cloud v2.0 — 5-Track Empirical Evaluation",
  "total_core_tasks": 57,
  "total_ood_holdout_tasks": 6,
  "tracks": 5,
  "verification_note": "Cryptographic SHA-256 digests of frozen benchmark suites, multi-iteration convergence telemetry, and unseen OOD holdout artifacts. Full execution harness and task fixtures are available to credentialed grant and technical due-diligence reviewers upon request (founder@thunderbearstudios.org).",
  "suite_digests": {
    "track1_baxbench_cweval_builder": {
      "tasks": 12,
      "paradigm": "ETH Zurich BaxBench / CWEval Dual Functional + Exploit Suite",
      "sha256": "13011631f64d71979a5fadeeffed42a6500b779a06a694e83f331e24c56a0cd9",
      "bytes": 333
    },
    "track1_baxbench_cweval_fixtures": {
      "tasks": 12,
      "paradigm": "10 MITRE CWE Families + Real-Time Audio + False-Positive Control",
      "sha256": "c05f91a75eb3456a7591649938e3695e2e3b6fbf9584ba8c07adcda01d59b9d2",
      "bytes": 40044
    },
    "track2_swebench_hal_builder": {
      "tasks": 8,
      "paradigm": "SWE-bench Verified Mini / Princeton HAL Multi-File Suite",
      "sha256": "47aa8f0d7be2d98ade370f269d3aa23609c99547d0b45f37d2ab72e8e45c6e6e",
      "bytes": 368
    },
    "track2_swebench_hal_fixtures": {
      "tasks": 8,
      "paradigm": "Cross-Caller Signature Sync, Dirty-State Gate, Cache Symmetry & Shadowing",
      "sha256": "83c25955c31ad6a042870d7f6aa1509b7c58bfcace54411b57fe48fb609c8fd4",
      "bytes": 40487
    },
    "tracks_1_2_ablation_runner": {
      "arms": 3,
      "sha256": "563afe72f2a43e668ecd2ee7eb87599ce4251d969398acba7e1b951a28ac1e3e",
      "bytes": 24322
    },
    "tracks_3_4_5_polyglot_camel_memory_suite": {
      "tasks": 37,
      "paradigm": "Multi-SWE-bench Polyglot + Compiled C++20/Rust RTSan + AgentDojo/CaMeL + LongMemEval/HippoRAG 2",
      "sha256": "eeca5ffdf41d26f11beb32132853755baca5c4ff73831a545e370c085499a277",
      "bytes": 60859
    }
  },
  "result_telemetry_digests": {
    "tracks_1_2_three_arm_report": {
      "secpass_at_1": "100.0% (12/12, p=0.000977)",
      "swebench_resolved": "100.0% (8/8, -93.76% context bytes, p=0.007812)",
      "sha256": "6bdea73b98e4f5412d79d4d14435cd4f4740ae1ffa38bfdebca9976514815776b",
      "bytes": 51102
    },
    "iteration_0_baseline_telemetry": {
      "composite_utility_pct": 8.33,
      "sha256": "35bb2d37cbae4ca121f739eb6550e2d04fa24129a8a3465810bd1978593a556a",
      "bytes": 26398
    },
    "iteration_1_telemetry": {
      "composite_utility_pct": 70.83,
      "sha256": "d0e5e02c6b4d1430eddbfff50c0689413c2193d4be5c063c5694e83e27db0c77",
      "bytes": 26398
    },
    "iteration_2_telemetry": {
      "composite_utility_pct": 98.33,
      "sha256": "d38c697bc6988d8060536d9fde03ed4835b33432ae3f7e6ce38c12356f7993a6",
      "bytes": 26393
    },
    "iteration_3_saturated_ceiling_telemetry": {
      "composite_utility_pct": 100.0,
      "sha256": "474083859d372eb35beb214df950e0b0a5d93acc76bc78e6602536834277b582",
      "bytes": 26393
    },
    "unseen_ood_holdout_post_freeze": {
      "holdout_pass_rate": "100.0% (6/6 unseen tasks)",
      "sha256": "09e454715790c18834d71877b5c9121be47d13bb0ca81b0fcda01d1c897863e0",
      "bytes": 27506
    },
    "convergence_checkpoint": {
      "sha256": "3f28c3b02179edf2671af5e39f3477005304c1611b1b98d4efe2353775fa8d35",
      "bytes": 1409
    }
  }
}
