{
  "version": "v1",
  "published": "2026-09-15",
  "model_family": "any chat-capable model (ollama, OpenAI, Anthropic, local)",
  "n_facts": 6,
  "protocol": "Meetless-style stale-context scoring. Each fact's CURRENT value is recoverable ONLY from its dated in-force note (the superseding note written last). Both arms carry the FULL dated-note history on disk (the Meetless gate: every answer is recoverable from every arm's corpus). The pushed arm additionally carries the operations bulletin: the in-force decision delivered completely, each naming the winning value (the AirMemo product mechanism: 'deliver the in-force decision into context, COMPLETELY, and say which value wins').",
  "scoring": "0..N, one point per question whose answer matches the in-force value (normalized: lowercase, punctuation stripped, alias map applies).",
  "questions": [
    {
      "id": "q1",
      "fact_id": "f1",
      "question": "What is the current maximum pod count for the web tier?",
      "answer": "12",
      "aliases": [
        "twelve pods",
        "12 pods"
      ]
    },
    {
      "id": "q2",
      "fact_id": "f2",
      "question": "What is the current deployment cadence for the payments service?",
      "answer": "weekly",
      "aliases": [
        "every week",
        "once a week"
      ]
    },
    {
      "id": "q3",
      "fact_id": "f3",
      "question": "Which database is the current primary for the accounting service?",
      "answer": "neon",
      "aliases": [
        "neon postgres",
        "neon db"
      ]
    },
    {
      "id": "q4",
      "fact_id": "f4",
      "question": "What is the current incident on-call rotation schedule?",
      "answer": "2-week",
      "aliases": [
        "biweekly",
        "two-week rotation",
        "every two weeks"
      ]
    },
    {
      "id": "q5",
      "fact_id": "f5",
      "question": "What is the current feature flag default for the billing v2 rollout?",
      "answer": "off",
      "aliases": [
        "disabled",
        "false",
        "flag off"
      ]
    },
    {
      "id": "q6",
      "fact_id": "f6",
      "question": "What is the current staging environment's upstream mirror source?",
      "answer": "production snapshots",
      "aliases": [
        "prod snapshots",
        "snapshots from prod"
      ]
    }
  ],
  "source": {
    "path": "benchmark/facts/v1.json",
    "sha256": "d629188fdfc129e3e07c5497f734c7c4be67b31a41bfd2d9290c1b9064e8acc7"
  }
}