{
  "name": "PactVerity AI Agent Evidence Benchmark",
  "version": "0.1",
  "published": "2026-08-21",
  "license": "CC-BY-4.0",
  "methodology": "https://pactverity.com/insights/ai-agent-evidence-benchmark-2026",
  "limitations": "First-party known-answer scenarios. The dataset tests deterministic receipt evaluation and does not establish independent monitoring, commercial adoption, settlement, audit assurance or token value.",
  "verifiedExternalRuns": 0,
  "scenarios": [
    {
      "id": "healthy-agent-response",
      "description": "All evaluated availability checks pass; output hash is intentionally not evaluated.",
      "expectedOutcome": "pass",
      "expectedCheckStatuses": ["pass", "pass", "pass", "not_evaluated"],
      "criteria": { "maxLatencyMs": 900, "minUptimeBps": 9950, "maxFreshnessSeconds": 120, "expectedOutputSha256": "" },
      "evidence": { "measuredLatencyMs": 642, "uptimeBps": 9988, "freshnessSeconds": 42, "observedOutputSha256": "" }
    },
    {
      "id": "latency-threshold-breach",
      "description": "Latency exceeds the agreement while uptime and freshness remain within bounds.",
      "expectedOutcome": "fail",
      "expectedCheckStatuses": ["fail", "pass", "pass", "not_evaluated"],
      "criteria": { "maxLatencyMs": 500, "minUptimeBps": 9900, "maxFreshnessSeconds": 180, "expectedOutputSha256": "" },
      "evidence": { "measuredLatencyMs": 781, "uptimeBps": 9975, "freshnessSeconds": 66, "observedOutputSha256": "" }
    },
    {
      "id": "output-integrity-mismatch",
      "description": "Availability checks pass but the observed output hash differs from the expected hash.",
      "expectedOutcome": "fail",
      "expectedCheckStatuses": ["pass", "pass", "pass", "fail"],
      "criteria": { "maxLatencyMs": 1000, "minUptimeBps": 9900, "maxFreshnessSeconds": 300, "expectedOutputSha256": "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" },
      "evidence": { "measuredLatencyMs": 410, "uptimeBps": 9990, "freshnessSeconds": 14, "observedOutputSha256": "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb" }
    }
  ]
}
