{
  "schema": "agentomy.benchmark-manifest/v1",
  "generated_from": {
    "receipt": ".ceremony-receipt.json (written by scripts/ceremony-rebuild-verify.sh after a full rebuild and run)",
    "canonical_record": "config/canonical-values.json (the published totals every surface is checked against)",
    "renderer": "scripts/render-benchmark-manifest.py",
    "gate": "scripts/verify-benchmark-manifest.py fails any commit whose served manifest is not this rendering, or whose receipt disagrees with the canonical record"
  },
  "verified_at": "2026-09-04T00:58:31Z",
  "results": {
    "governancebench_core": {
      "what": "The core conformance set every implementation is held to: the scenarios a governance layer must catch, scored across six dimensions",
      "scenarios": 235,
      "passed": 235,
      "failed": 0,
      "score": "100/100",
      "dimensions": "6/6",
      "runner": {
        "package": "governancebench",
        "version": "1.3.5",
        "registry": "npm"
      },
      "scenario_manifest": {
        "files": 21,
        "sha256": "ff5e3abd1628d10a1c3a2df76562253de01106ac5f5e184ecfe0c167355d83f7",
        "pattern": "cli/governancebench/suites/*.mjs"
      }
    },
    "governancebench_optional": {
      "what": "Optional capability suites for verticals and patterns, scored standalone; not part of the core figure",
      "scenarios": 218,
      "suites": 21,
      "all_suites_total": 453
    },
    "vigil": {
      "what": "Adversarial testing: attack and failure scenarios across the threat categories, run against the same target",
      "scenarios": 148,
      "passed": 148,
      "failed": 0,
      "score": "100/100",
      "categories": 14,
      "runner": {
        "package": "agentomy-vigil",
        "version": "1.0.3",
        "registry": "npm"
      },
      "scenario_manifest": {
        "files": 14,
        "sha256": "df86c632566b776b112558b0c60a89287922fbe62bdbef1b28c3b56bc2c69b87",
        "pattern": "cli/vigil/categories/*.mjs"
      }
    },
    "runtime_tests": {
      "what": "The platform's own test suite, run in the same ceremony before the container is rebuilt",
      "tests": 9339,
      "passed": 9339,
      "suites": 326
    },
    "kevlar": {
      "what": "OWASP-aligned adversarial scoring by an external evaluator; the one externally verified figure, recorded as published",
      "scenarios": 48,
      "score": "10/10"
    }
  },
  "scored_files": {
    "what": "Every file whose hash the receipt binds to the figures above; a change to any of them re-runs the ceremony before it can be committed",
    "count": 58,
    "sha256_of_hash_map": "d6c49b329a731cc1164dcf68c0574fce6256040e8e812737f64d68f955974665"
  },
  "result_files": [
    {
      "path": "governancebench-results/results.json",
      "present": true,
      "description": "GovernanceBench core run: the scored scenarios with pass, reason and elapsed time per scenario",
      "generated_at": "2026-08-10T07:37:46.583Z",
      "summary": {
        "total": 232,
        "passed": 232,
        "failed": 0,
        "skipped": 0
      },
      "format_version": "1.0.0"
    },
    {
      "path": "governancebench-results/message-governance-results.json",
      "present": true,
      "description": "GovernanceBench message-governance suite run",
      "generated_at": "2026-08-10T07:10:52.558Z",
      "summary": {
        "total": 8,
        "passed": 8,
        "failed": 0,
        "skipped": 0
      },
      "format_version": "1.0.0"
    },
    {
      "path": "governancebench-results/vigil/results.json",
      "present": true,
      "description": "VIGIL run: every category and scenario with its result; rewritten by every ceremony",
      "generated_at": "2026-09-04T00:58:29.953Z",
      "summary": {
        "total": 148,
        "passed": 148,
        "failed": 0,
        "skipped": 0
      },
      "format_version": "1.0.0"
    },
    {
      "path": "governancebench-results/workflowbench/agentomy-results.json",
      "present": true,
      "description": "WorkflowBench run against the Agentomy adapter",
      "generated_at": "2026-08-10T07:10:52.194Z",
      "benchmark": "WorkflowBench",
      "version": "0.1.0"
    }
  ],
  "reproduce": {
    "governancebench": "npx governancebench run --target <your platform> --suite all --api-key <key>",
    "vigil": "npx agentomy-vigil run --target <your platform> --api-key <key>",
    "note": "The runners accept any target that speaks the governance API; the figures above are what they report against Agentomy. Run them against your own deployment and compare."
  },
  "links": {
    "governancebench": "https://agentomy.com/governancebench",
    "evaluator_guide": "https://agentomy.com/docs",
    "changes_feed": "https://agentomy.com/changes.json",
    "service": "https://agentomy.com/service.json"
  }
}