{
  "format": "orbit-benchmark-results-v1",
  "suite": "A",
  "executedAt": "2026-09-29T19:47:00.953024-04:00",
  "corpusStatus": "development-not-frozen",
  "engineSha256": "d0f5a482ed1ce1f81abf991d7a028328a25a73236a9e967e61de656a3fe51f0a",
  "relationEngineSha256": "d5da24352b213c55cede9559e2faa48b3b8451c57a3257b882c77ff82ee495a4",
  "oracleSha256": "aabe0f8b121f32a5eda7795dfe730f70993829a30b18c1857974b5cdf4e6fcd0",
  "summary": {
    "baseline": {
      "questions": 36,
      "correct": 36,
      "invariantsPassed": 108,
      "invariantsTotal": 108,
      "relationPassed": 18,
      "relationTotal": 18
    },
    "n": {
      "questions": 36,
      "correct": 36,
      "invariantsPassed": 108,
      "invariantsTotal": 108,
      "relationPassed": 18,
      "relationTotal": 18
    },
    "p": {
      "questions": 36,
      "correct": 36,
      "invariantsPassed": 108,
      "invariantsTotal": 108,
      "relationPassed": 18,
      "relationTotal": 18
    }
  },
  "limitations": [
    "Development synthetic corpus, not a held-out final evaluation.",
    "Baseline and N intentionally share the same admissibility controls.",
    "No autonomous model or WebMCP agent result is represented here."
  ]
}
