{
  "title": "8DB application operations: inspected implementation and evaluation scope",
  "reviewed_at": "2026-09-24",
  "evidence_class": "source-inspected; no fresh engine execution",
  "source_revision": "7a3654189fb29a1ddf26843575ee9749ab5c7243",
  "findings": [
    {
      "id": "D01",
      "source": "time_series_carrier.rs; time_series_absence.rs",
      "claim": "Typed gap classifications distinguish reingestion, awaiting data and not-applicable intervals.",
      "scope": "Cadence/deployment context is separate from transaction-version observation filtering."
    },
    {
      "id": "D02",
      "source": "scientific_lineage.rs",
      "claim": "BioTwin separates content and lineage fingerprints and queues affected work for rerun.",
      "scope": "Application-specific source review; multi-process and recovered-history coverage require qualification."
    },
    {
      "id": "D03",
      "source": "solution_space.rs",
      "claim": "BioTwin six-axis ordering calls shared lattice operations.",
      "scope": "Fixed-point score input, missing-value and rounding policies affect the decision."
    },
    {
      "id": "D06",
      "source": "channel.rs",
      "claim": "Message body encrypted once; content key wrapped for current recipients.",
      "scope": "Membership agreement, restart and replay checks are separate from the qualified lexical exchange profile."
    },
    {
      "id": "D07",
      "source": "vault/store/mod.rs",
      "claim": "Durable Vault consumer uses encrypted chunks and sealed manifests with a whole-blob fallback.",
      "scope": "Chunks and manifests commit separately. Measure full stored bytes and interrupted-write recovery; historical in-memory ratios do not transfer."
    },
    {
      "id": "D08",
      "source": "time_series_layers.rs",
      "claim": "Selected coarse view carries residuals, staleness and confidence.",
      "scope": "Projection rebuilt from fine samples per call; coarse preview is not an exact fine alarm or reduced-read benchmark."
    }
  ]
}