{
  "schema": "wattsutra.audit.v1",
  "toolVersion": "0.2.1",
  "generatedAt": "2026-11-16T17:56:27.058Z",
  "environment": {
    "node": "v24.15.0",
    "platform": "win32",
    "arch": "x64"
  },
  "verdict": "PASS",
  "scope": "Evidence integrity, claim hygiene, comparison fairness and packaging. Energy, carbon and cost gates are NOT_RUN: no hardware meter was used.",
  "totals": {
    "checks": 18,
    "verified": 17,
    "failed": 0,
    "unknown": 0,
    "notRun": 1,
    "failedBySeverity": {
      "critical": 0,
      "high": 0,
      "medium": 0,
      "low": 0,
      "info": 0
    }
  },
  "checks": [
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A01",
      "title": "Benchmark runs verify and keep quality parity",
      "status": "verified",
      "evidence": "baseline 9a1da4495e209034f9ee4c77481ef39150d65938e70d0bf8a36d6114c0d904b9 candidate 07d0c0ac64c6fa76987341a166ef263b1fbb18c3f6bf4aeb61cfd76a1819f2e3; prime count 2262, sum 21171191",
      "command": "wattsutra bench --manifest fixtures/baseline.json",
      "risk": "A faster variant that computes a different answer would be reported as an improvement",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A02",
      "title": "Edited reports are rejected (integrity hash)",
      "status": "verified",
      "evidence": "mutated wallMs.mean; verifyRun threw integrity mismatch",
      "command": "node --test tests/",
      "risk": "Silent result tampering",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A03",
      "title": "Internally inconsistent reports are rejected even with a recomputed hash",
      "status": "verified",
      "evidence": "mean edited, hash recomputed; validateRunShape recomputed stats from samples and refused",
      "command": "node --test tests/",
      "risk": "Hash alone proves only no later edits, not honest content. A fully re-derived forgery (samples and hash) is NOT detectable without signing",
      "action": "Optional Sigstore signing is on the roadmap"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A04",
      "title": "Measured runs contain no energy, carbon or cost values",
      "status": "verified",
      "evidence": "energyKWh/carbonGCO2e/cost are null; injected energyKWh=0.5 was refused",
      "command": "wattsutra bench",
      "risk": "CPU time presented as joules or CO2e",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A05",
      "title": "Incomparable workloads are refused",
      "status": "verified",
      "evidence": "n=20000 vs n=3000 refused",
      "command": "wattsutra compare",
      "risk": "Apples-to-oranges speedup claims",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A06",
      "title": "Quality mismatch yields a FAIL gate",
      "status": "verified",
      "evidence": "gate=FAIL",
      "command": "wattsutra compare",
      "risk": "Speedup accepted although answers differ",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A07",
      "title": "Environment mismatch downgrades the gate to REVIEW_REQUIRED",
      "status": "verified",
      "evidence": "mismatched: cpuModel",
      "command": "wattsutra compare",
      "risk": "Cross-machine numbers shown as a like-for-like result",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A08",
      "title": "Too few samples never earns PASS",
      "status": "verified",
      "evidence": "n=2 per run, gate=REVIEW_REQUIRED",
      "command": "wattsutra compare",
      "risk": "Tiny samples described as conclusive",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A09",
      "title": "Estimator applies no default rates",
      "status": "verified",
      "evidence": "unknown: energyKWh, operationalGCO2e, cost",
      "command": "wattsutra estimate (empty assumptions)",
      "risk": "Invented watts, carbon intensity or prices",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A10",
      "title": "Estimator rejects unknown fields and ambiguous energy input",
      "status": "verified",
      "evidence": "typo key and dual energy inputs refused",
      "command": "wattsutra estimate",
      "risk": "Misspelled assumption silently ignored",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A11",
      "title": "Modeled and imported values are labelled",
      "status": "verified",
      "evidence": "kinds: modeled, imported-meter-observation (attributionUnverified=true)",
      "command": "wattsutra estimate",
      "risk": "Measured vs modeled confusion",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A12",
      "title": "HTML report escapes hostile content",
      "status": "verified",
      "evidence": "payload escaped; CSP meta present: true",
      "command": "wattsutra report --format html",
      "risk": "Script injection through report fields",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A13",
      "title": "Hashing is deterministic regardless of key order",
      "status": "verified",
      "evidence": "canonical JSON, code-unit key order",
      "command": "node --test tests/",
      "risk": "Hash differs across machines or locales",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A14",
      "title": "Skills and agent manifest are consistent",
      "status": "verified",
      "evidence": "20 skills, 20 manifest entries",
      "command": "wattsutra audit",
      "risk": "Skill discovery fails silently in IDEs",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "security",
      "id": "A15",
      "title": "No credentials in shipped files",
      "status": "verified",
      "evidence": "regex scan for GitHub, npm, AWS, OpenAI-style keys and private keys: no matches",
      "command": "wattsutra audit",
      "risk": "Leaked secrets in a public package",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A16",
      "title": "Installer refuses to overwrite customised skills",
      "status": "verified",
      "evidence": "second install refused to clobber edited SKILL.md",
      "command": "wattsutra install",
      "risk": "User edits destroyed",
      "action": "None"
    },
    {
      "severity": "info",
      "owner": "maintainer",
      "id": "A17",
      "title": "Hardware energy instrument",
      "status": "not-run",
      "evidence": "RAPL unavailable; tools: nvidia-smi=unavailable, powermetrics=unavailable, scaphandre=unavailable; measuredEnergyAvailable=false",
      "command": "wattsutra doctor",
      "risk": "Without a meter, energy and carbon stay unmeasured. This is deliberate and correct.",
      "action": "Import an external meter reading via estimate --assumptions to get an imported-meter-observation"
    },
    {
      "severity": "low",
      "owner": "user",
      "id": "A18",
      "title": "Statistical confidence",
      "status": "verified",
      "evidence": "latency gate PASS; n=12/12; ratio 4.66x; 95% bootstrap interval [3.46, 5.34]; CV baseline 42.6%",
      "command": "wattsutra compare",
      "risk": "Small samples on a shared machine; interval is bootstrap, not a significance test",
      "action": "Repeat on a quiet, pinned machine for publication-grade claims"
    }
  ],
  "sample": {
    "baseline": {
      "schema": "wattsutra.run.v1",
      "product": "WattSutra AI",
      "toolVersion": "0.2.1",
      "workloadId": "prime-sum",
      "manifestHash": "843ad33c3ba0960a39829841c69d16b337d05906900c84c349bc4ffa8ff460a1",
      "workload": {
        "n": 20000,
        "variant": "trial",
        "functionalUnit": "one complete prime-sum job",
        "checksum": "114079488afea6df6c74454f33b2063639d5737cd34b1e1e7d551dd12ec181ed"
      },
      "environment": {
        "platform": "win32",
        "arch": "x64",
        "node": "v24.15.0",
        "cpuModel": "Intel(R) Core(TM) Ultra 7 155U",
        "logicalCores": 14,
        "osRelease": "10.0.26300",
        "totalMemBytes": 33767481344,
        "loadAvg1m": null
      },
      "methodology": {
        "version": "offline-probe-v1",
        "warmups": 1,
        "qualityCheck": "reference trial prime count and sum",
        "sampleScope": "single process, sequential",
        "rssScope": "whole Node.js process (not incremental workload bytes)",
        "cpuScope": "process cpuUsage for measured interval (not system energy)",
        "equipment": "no power meter",
        "clock": "process.hrtime.bigint (monotonic)",
        "confidence": "timing and process CPU sampled; memory process-wide"
      },
      "measurements": {
        "wallMs": {
          "kind": "measured",
          "n": 12,
          "min": 0.902,
          "max": 2.6897,
          "mean": 1.1852916666666669,
          "median": 0.96275,
          "p95": 1.9823999999999993,
          "stddev": 0.5050529899441396,
          "stddevKind": "sample (n-1)",
          "cv": 0.426100177827516,
          "unit": "ms",
          "samples": [
            0.9607,
            0.9138,
            0.902,
            0.9371,
            0.98,
            0.9648,
            0.9496,
            0.9404,
            1.4037,
            1.304,
            1.2777,
            2.6897
          ]
        },
        "cpuMs": {
          "kind": "measured",
          "n": 12,
          "min": 0,
          "max": 16,
          "mean": 2.5833333333333335,
          "median": 0,
          "p95": 15.45,
          "stddev": 6.0371325724196,
          "stddevKind": "sample (n-1)",
          "cv": 2.3369545441624258,
          "unit": "ms",
          "samples": [
            15,
            16,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0
          ]
        },
        "rssBytes": {
          "kind": "measured",
          "n": 12,
          "min": 64327680,
          "max": 64671744,
          "mean": 64465237.333333336,
          "median": 64430080,
          "p95": 64662732.8,
          "stddev": 130914.32940899195,
          "stddevKind": "sample (n-1)",
          "cv": 0.0020307740237124586,
          "unit": "bytes",
          "samples": [
            64327680,
            64327680,
            64331776,
            64372736,
            64380928,
            64397312,
            64462848,
            64487424,
            64532480,
            64634880,
            64655360,
            64671744
          ]
        }
      },
      "unmeasured": [
        "GPU energy",
        "CPU electrical power",
        "electricity kWh",
        "CO2e",
        "cloud billing cost",
        "end-to-end inference quality"
      ],
      "energyKWh": null,
      "carbonGCO2e": null,
      "cost": null,
      "quality": {
        "kind": "verified",
        "reference": {
          "count": 2262,
          "sum": 21171191
        },
        "actual": {
          "count": 2262,
          "sum": 21171191
        }
      },
      "recordedAt": "2026-11-16T17:56:26.426Z",
      "runId": "run-4f8fc2245e40cb96",
      "integrity": "9a1da4495e209034f9ee4c77481ef39150d65938e70d0bf8a36d6114c0d904b9"
    },
    "candidate": {
      "schema": "wattsutra.run.v1",
      "product": "WattSutra AI",
      "toolVersion": "0.2.1",
      "workloadId": "prime-sum",
      "manifestHash": "c0fb95217b792abc0cdb2ac89ca603398d222cb294e6b662a1bbdf0f65add389",
      "workload": {
        "n": 20000,
        "variant": "sieve",
        "functionalUnit": "one complete prime-sum job",
        "checksum": "114079488afea6df6c74454f33b2063639d5737cd34b1e1e7d551dd12ec181ed"
      },
      "environment": {
        "platform": "win32",
        "arch": "x64",
        "node": "v24.15.0",
        "cpuModel": "Intel(R) Core(TM) Ultra 7 155U",
        "logicalCores": 14,
        "osRelease": "10.0.26300",
        "totalMemBytes": 33767481344,
        "loadAvg1m": null
      },
      "methodology": {
        "version": "offline-probe-v1",
        "warmups": 1,
        "qualityCheck": "reference trial prime count and sum",
        "sampleScope": "single process, sequential",
        "rssScope": "whole Node.js process (not incremental workload bytes)",
        "cpuScope": "process cpuUsage for measured interval (not system energy)",
        "equipment": "no power meter",
        "clock": "process.hrtime.bigint (monotonic)",
        "confidence": "timing and process CPU sampled; memory process-wide"
      },
      "measurements": {
        "wallMs": {
          "kind": "measured",
          "n": 12,
          "min": 0.2099,
          "max": 0.335,
          "mean": 0.2542833333333333,
          "median": 0.24409999999999998,
          "p95": 0.311955,
          "stddev": 0.03669807607871002,
          "stddevKind": "sample (n-1)",
          "cv": 0.14431962802140666,
          "unit": "ms",
          "samples": [
            0.335,
            0.2358,
            0.2931,
            0.2171,
            0.2742,
            0.2239,
            0.2099,
            0.2263,
            0.2457,
            0.2761,
            0.2425,
            0.2718
          ]
        },
        "cpuMs": {
          "kind": "measured",
          "n": 12,
          "min": 0,
          "max": 0,
          "mean": 0,
          "median": 0,
          "p95": 0,
          "stddev": 0,
          "stddevKind": "sample (n-1)",
          "cv": null,
          "unit": "ms",
          "samples": [
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0
          ]
        },
        "rssBytes": {
          "kind": "measured",
          "n": 12,
          "min": 65601536,
          "max": 66064384,
          "mean": 65757525.333333336,
          "median": 65703936,
          "p95": 66044108.8,
          "stddev": 169607.4807726965,
          "stddevKind": "sample (n-1)",
          "cv": 0.0025792862476641935,
          "unit": "bytes",
          "samples": [
            65601536,
            65601536,
            65601536,
            65609728,
            65638400,
            65687552,
            65720320,
            65781760,
            65830912,
            65925120,
            66027520,
            66064384
          ]
        }
      },
      "unmeasured": [
        "GPU energy",
        "CPU electrical power",
        "electricity kWh",
        "CO2e",
        "cloud billing cost",
        "end-to-end inference quality"
      ],
      "energyKWh": null,
      "carbonGCO2e": null,
      "cost": null,
      "quality": {
        "kind": "verified",
        "reference": {
          "count": 2262,
          "sum": 21171191
        },
        "actual": {
          "count": 2262,
          "sum": 21171191
        }
      },
      "recordedAt": "2026-11-16T17:56:26.439Z",
      "runId": "run-0b32abe9c0858b0c",
      "integrity": "07d0c0ac64c6fa76987341a166ef263b1fbb18c3f6bf4aeb61cfd76a1819f2e3"
    },
    "comparison": {
      "schema": "wattsutra.comparison.v1",
      "toolVersion": "0.2.1",
      "baseline": {
        "runId": "run-4f8fc2245e40cb96",
        "variant": "trial",
        "hash": "9a1da4495e209034f9ee4c77481ef39150d65938e70d0bf8a36d6114c0d904b9",
        "meanMs": 1.1852916666666669,
        "samples": 12
      },
      "candidate": {
        "runId": "run-0b32abe9c0858b0c",
        "variant": "sieve",
        "hash": "07d0c0ac64c6fa76987341a166ef263b1fbb18c3f6bf4aeb61cfd76a1819f2e3",
        "meanMs": 0.2542833333333333,
        "samples": 12
      },
      "sameWorkload": true,
      "qualityEquivalent": true,
      "environmentParity": {
        "matched": true,
        "mismatchedFields": []
      },
      "latencyRatio": 4.661303008455136,
      "latencyRatioMedian": 3.944080294961082,
      "latencyChangePercent": -78.54677118852604,
      "latencyRatioInterval95": {
        "low": 3.459896983075791,
        "high": 5.342072920934044,
        "resamples": 2000,
        "seed": 359031,
        "method": "percentile bootstrap of median(baseline)/median(candidate); medians resist outliers such as GC pauses"
      },
      "gates": {
        "latency": "PASS",
        "energy": "NOT_RUN",
        "carbon": "NOT_RUN",
        "cost": "NOT_RUN"
      },
      "gateReasons": [
        "Candidate faster with equal quality, matched environment and an interval that excludes no change"
      ],
      "measurementOnly": true,
      "energySaving": null,
      "carbonSaving": null,
      "costSaving": null,
      "notes": [
        "Differences reflect these local benchmark runs, not guaranteed future performance",
        "No electrical power, carbon, or cost saving is inferred from latency alone",
        "Energy, carbon and cost gates are NOT_RUN: no meter or user-supplied model was provided to this comparison"
      ]
    },
    "estimate": {
      "schema": "wattsutra.estimate.v1",
      "toolVersion": "0.2.1",
      "sourceHash": "07d0c0ac64c6fa76987341a166ef263b1fbb18c3f6bf4aeb61cfd76a1819f2e3",
      "method": "explicit-inputs-only",
      "estimated": {
        "energyKWh": {
          "value": 4.591226851851852e-9,
          "kind": "modeled",
          "formula": "user-supplied averagePowerW × observed wall-hours / 1000",
          "boundary": "declared power scope; may not equal process energy"
        },
        "operationalGCO2e": {
          "value": 0.0000017446662037037039,
          "kind": "modeled",
          "formula": "energyKWh × user-provided carbonIntensityGPerKWh"
        },
        "cost": {
          "value": 0.002,
          "kind": "modeled",
          "unit": "USD/job",
          "source": "user-supplied price; not observed billing"
        }
      },
      "observed": {},
      "unknown": [],
      "assumptions": {
        "averagePowerW": 65,
        "carbonIntensityGPerKWh": 380,
        "ratePerJob": 0.002,
        "currency": "USD"
      },
      "caveats": [
        "No default electricity, carbon or pricing rates are applied",
        "Process elapsed time cannot directly measure electrical energy",
        "SCI score not claimed without full boundary, functional unit and embodied methodology"
      ]
    }
  }
}
