{
  "ok": true,
  "cases": "evals/output/cases.jsonl",
  "summary": {
    "case_count": 5,
    "file_backed_case_count": 1,
    "near_neighbor_case_count": 1,
    "boundary_case_count": 1,
    "baseline_pass_rate": 0.0,
    "with_skill_pass_rate": 100.0,
    "delta": 100.0,
    "regression_count": 0,
    "gate_pass": true,
    "failure_taxonomy": [],
    "blind_pair_count": 5
  },
  "results": [
    {
      "id": "skill-package-contract",
      "prompt": "Turn this repeated workflow into a reusable team skill package.",
      "input_files": [],
      "metadata": {},
      "baseline": {
        "score": 0.0,
        "passed_count": 0,
        "failed_count": 4,
        "checks": [
          {
            "id": "has-entrypoint",
            "description": "Output names the routeable Skill entrypoint.",
            "weight": 1.0,
            "failure_type": "missing_entrypoint",
            "passed": false,
            "missing": [
              "SKILL.md"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-interface",
            "description": "Output includes neutral interface metadata.",
            "weight": 1.0,
            "failure_type": "missing_interface",
            "passed": false,
            "missing": [
              "agents/interface.yaml"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-report",
            "description": "Output points reviewer to generated overview report.",
            "weight": 1.0,
            "failure_type": "missing_review_artifact",
            "passed": false,
            "missing": [
              "reports/skill-overview.html"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-resource-boundary",
            "description": "Output preserves entrypoint/resource separation.",
            "weight": 1.0,
            "failure_type": "resource_boundary",
            "passed": false,
            "missing": [
              "references/",
              "scripts/",
              "reports/"
            ],
            "present_forbidden": []
          }
        ],
        "failed": [
          {
            "id": "has-entrypoint",
            "description": "Output names the routeable Skill entrypoint.",
            "weight": 1.0,
            "failure_type": "missing_entrypoint",
            "passed": false,
            "missing": [
              "SKILL.md"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-interface",
            "description": "Output includes neutral interface metadata.",
            "weight": 1.0,
            "failure_type": "missing_interface",
            "passed": false,
            "missing": [
              "agents/interface.yaml"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-report",
            "description": "Output points reviewer to generated overview report.",
            "weight": 1.0,
            "failure_type": "missing_review_artifact",
            "passed": false,
            "missing": [
              "reports/skill-overview.html"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-resource-boundary",
            "description": "Output preserves entrypoint/resource separation.",
            "weight": 1.0,
            "failure_type": "resource_boundary",
            "passed": false,
            "missing": [
              "references/",
              "scripts/",
              "reports/"
            ],
            "present_forbidden": []
          }
        ]
      },
      "with_skill": {
        "score": 100.0,
        "passed_count": 4,
        "failed_count": 0,
        "checks": [
          {
            "id": "has-entrypoint",
            "description": "Output names the routeable Skill entrypoint.",
            "weight": 1.0,
            "failure_type": "missing_entrypoint",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-interface",
            "description": "Output includes neutral interface metadata.",
            "weight": 1.0,
            "failure_type": "missing_interface",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-report",
            "description": "Output points reviewer to generated overview report.",
            "weight": 1.0,
            "failure_type": "missing_review_artifact",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-resource-boundary",
            "description": "Output preserves entrypoint/resource separation.",
            "weight": 1.0,
            "failure_type": "resource_boundary",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          }
        ],
        "failed": []
      },
      "delta": 100.0,
      "winner": "with_skill",
      "failure_taxonomy": []
    },
    {
      "id": "output-eval-expectation",
      "prompt": "Upgrade this production skill so we know whether the generated output is better than baseline.",
      "input_files": [],
      "metadata": {},
      "baseline": {
        "score": 0.0,
        "passed_count": 0,
        "failed_count": 4,
        "checks": [
          {
            "id": "has-baseline-comparison",
            "description": "Output explicitly compares with-skill and baseline outputs.",
            "weight": 1.0,
            "failure_type": "missing_baseline_delta",
            "passed": false,
            "missing": [
              "with-skill",
              "baseline"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-assertion-grading",
            "description": "Output includes objective assertion grading.",
            "weight": 1.0,
            "failure_type": "missing_assertions",
            "passed": false,
            "missing": [
              "assertion grading"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-scorecard",
            "description": "Output produces a scorecard report path.",
            "weight": 1.0,
            "failure_type": "missing_scorecard",
            "passed": false,
            "missing": [
              "reports/output_quality_scorecard.md"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-failure-taxonomy",
            "description": "Output records failure taxonomy.",
            "weight": 1.0,
            "failure_type": "missing_failure_taxonomy",
            "passed": false,
            "missing": [
              "failure taxonomy"
            ],
            "present_forbidden": []
          }
        ],
        "failed": [
          {
            "id": "has-baseline-comparison",
            "description": "Output explicitly compares with-skill and baseline outputs.",
            "weight": 1.0,
            "failure_type": "missing_baseline_delta",
            "passed": false,
            "missing": [
              "with-skill",
              "baseline"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-assertion-grading",
            "description": "Output includes objective assertion grading.",
            "weight": 1.0,
            "failure_type": "missing_assertions",
            "passed": false,
            "missing": [
              "assertion grading"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-scorecard",
            "description": "Output produces a scorecard report path.",
            "weight": 1.0,
            "failure_type": "missing_scorecard",
            "passed": false,
            "missing": [
              "reports/output_quality_scorecard.md"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-failure-taxonomy",
            "description": "Output records failure taxonomy.",
            "weight": 1.0,
            "failure_type": "missing_failure_taxonomy",
            "passed": false,
            "missing": [
              "failure taxonomy"
            ],
            "present_forbidden": []
          }
        ]
      },
      "with_skill": {
        "score": 100.0,
        "passed_count": 4,
        "failed_count": 0,
        "checks": [
          {
            "id": "has-baseline-comparison",
            "description": "Output explicitly compares with-skill and baseline outputs.",
            "weight": 1.0,
            "failure_type": "missing_baseline_delta",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-assertion-grading",
            "description": "Output includes objective assertion grading.",
            "weight": 1.0,
            "failure_type": "missing_assertions",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-scorecard",
            "description": "Output produces a scorecard report path.",
            "weight": 1.0,
            "failure_type": "missing_scorecard",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-failure-taxonomy",
            "description": "Output records failure taxonomy.",
            "weight": 1.0,
            "failure_type": "missing_failure_taxonomy",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          }
        ],
        "failed": []
      },
      "delta": 100.0,
      "winner": "with_skill",
      "failure_taxonomy": []
    },
    {
      "id": "ir-before-packaging",
      "prompt": "Package this skill for OpenAI, Claude, Agent Skills, and generic targets.",
      "input_files": [],
      "metadata": {},
      "baseline": {
        "score": 0.0,
        "passed_count": 0,
        "failed_count": 4,
        "checks": [
          {
            "id": "has-ir-first",
            "description": "Output requires Skill IR before packaging.",
            "weight": 1.0,
            "failure_type": "missing_ir",
            "passed": false,
            "missing": [
              "Skill IR first"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-core-ir-fields",
            "description": "Output lists core IR contract fields.",
            "weight": 1.0,
            "failure_type": "incomplete_ir_contract",
            "passed": false,
            "missing": [
              "job_to_be_done",
              "trigger_surface",
              "eval_plan",
              "governance"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-targets",
            "description": "Output names requested runtime targets.",
            "weight": 1.0,
            "failure_type": "missing_targets",
            "passed": false,
            "missing": [
              "OpenAI",
              "Claude",
              "Agent Skills",
              "generic"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-semantic-preservation",
            "description": "Output says targets must preserve the capability contract.",
            "weight": 1.0,
            "failure_type": "semantic_drift",
            "passed": false,
            "missing": [
              "preserve the same capability contract"
            ],
            "present_forbidden": []
          }
        ],
        "failed": [
          {
            "id": "has-ir-first",
            "description": "Output requires Skill IR before packaging.",
            "weight": 1.0,
            "failure_type": "missing_ir",
            "passed": false,
            "missing": [
              "Skill IR first"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-core-ir-fields",
            "description": "Output lists core IR contract fields.",
            "weight": 1.0,
            "failure_type": "incomplete_ir_contract",
            "passed": false,
            "missing": [
              "job_to_be_done",
              "trigger_surface",
              "eval_plan",
              "governance"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-targets",
            "description": "Output names requested runtime targets.",
            "weight": 1.0,
            "failure_type": "missing_targets",
            "passed": false,
            "missing": [
              "OpenAI",
              "Claude",
              "Agent Skills",
              "generic"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-semantic-preservation",
            "description": "Output says targets must preserve the capability contract.",
            "weight": 1.0,
            "failure_type": "semantic_drift",
            "passed": false,
            "missing": [
              "preserve the same capability contract"
            ],
            "present_forbidden": []
          }
        ]
      },
      "with_skill": {
        "score": 100.0,
        "passed_count": 4,
        "failed_count": 0,
        "checks": [
          {
            "id": "has-ir-first",
            "description": "Output requires Skill IR before packaging.",
            "weight": 1.0,
            "failure_type": "missing_ir",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-core-ir-fields",
            "description": "Output lists core IR contract fields.",
            "weight": 1.0,
            "failure_type": "incomplete_ir_contract",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-targets",
            "description": "Output names requested runtime targets.",
            "weight": 1.0,
            "failure_type": "missing_targets",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-semantic-preservation",
            "description": "Output says targets must preserve the capability contract.",
            "weight": 1.0,
            "failure_type": "semantic_drift",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          }
        ],
        "failed": []
      },
      "delta": 100.0,
      "winner": "with_skill",
      "failure_taxonomy": []
    },
    {
      "id": "near-neighbor-boundary",
      "prompt": "I only need a one-off summary of these notes, not a reusable process.",
      "input_files": [],
      "metadata": {
        "case_type": "near_neighbor",
        "tier": "production"
      },
      "baseline": {
        "score": 0.0,
        "passed_count": 0,
        "failed_count": 3,
        "checks": [
          {
            "id": "declines-skill",
            "description": "Output refuses unnecessary skill creation for one-off work.",
            "weight": 1.0,
            "failure_type": "over_trigger",
            "passed": false,
            "missing": [
              "Do not create a skill"
            ],
            "present_forbidden": []
          },
          {
            "id": "names-near-neighbor",
            "description": "Output labels the request as near-neighbor instead of owned work.",
            "weight": 1.0,
            "failure_type": "missing_boundary_language",
            "passed": false,
            "missing": [
              "near-neighbor"
            ],
            "present_forbidden": []
          },
          {
            "id": "requires-reuse-signal",
            "description": "Output asks for repeat-use evidence before packaging.",
            "weight": 1.0,
            "failure_type": "missing_reuse_gate",
            "passed": false,
            "missing": [
              "repeated use",
              "reusable output contract"
            ],
            "present_forbidden": []
          }
        ],
        "failed": [
          {
            "id": "declines-skill",
            "description": "Output refuses unnecessary skill creation for one-off work.",
            "weight": 1.0,
            "failure_type": "over_trigger",
            "passed": false,
            "missing": [
              "Do not create a skill"
            ],
            "present_forbidden": []
          },
          {
            "id": "names-near-neighbor",
            "description": "Output labels the request as near-neighbor instead of owned work.",
            "weight": 1.0,
            "failure_type": "missing_boundary_language",
            "passed": false,
            "missing": [
              "near-neighbor"
            ],
            "present_forbidden": []
          },
          {
            "id": "requires-reuse-signal",
            "description": "Output asks for repeat-use evidence before packaging.",
            "weight": 1.0,
            "failure_type": "missing_reuse_gate",
            "passed": false,
            "missing": [
              "repeated use",
              "reusable output contract"
            ],
            "present_forbidden": []
          }
        ]
      },
      "with_skill": {
        "score": 100.0,
        "passed_count": 3,
        "failed_count": 0,
        "checks": [
          {
            "id": "declines-skill",
            "description": "Output refuses unnecessary skill creation for one-off work.",
            "weight": 1.0,
            "failure_type": "over_trigger",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "names-near-neighbor",
            "description": "Output labels the request as near-neighbor instead of owned work.",
            "weight": 1.0,
            "failure_type": "missing_boundary_language",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "requires-reuse-signal",
            "description": "Output asks for repeat-use evidence before packaging.",
            "weight": 1.0,
            "failure_type": "missing_reuse_gate",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          }
        ],
        "failed": []
      },
      "delta": 100.0,
      "winner": "with_skill",
      "failure_taxonomy": []
    },
    {
      "id": "file-backed-governed-package",
      "prompt": "Turn the attached release brief source into a governed skill package.",
      "input_files": [
        "fixtures/release-brief-source.md"
      ],
      "metadata": {
        "case_type": "boundary",
        "tier": "governed"
      },
      "baseline": {
        "score": 0.0,
        "passed_count": 0,
        "failed_count": 5,
        "checks": [
          {
            "id": "uses-file-backed-evidence",
            "description": "Output names file-backed source evidence.",
            "weight": 1.0,
            "failure_type": "missing_file_fixture",
            "passed": false,
            "missing": [
              "file-backed fixture",
              "input_files"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-governance",
            "description": "Output preserves governed ownership metadata.",
            "weight": 1.0,
            "failure_type": "missing_governance",
            "passed": false,
            "missing": [
              "review cadence"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-output-contract",
            "description": "Output preserves output and rollback boundaries.",
            "weight": 1.0,
            "failure_type": "missing_contract_boundary",
            "passed": false,
            "missing": [
              "output contract",
              "rollback boundary"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-trust-and-scorecard",
            "description": "Output requires trust and output scorecard artifacts.",
            "weight": 1.0,
            "failure_type": "missing_release_evidence",
            "passed": false,
            "missing": [
              "trust report",
              "reports/output_quality_scorecard.md"
            ],
            "present_forbidden": []
          },
          {
            "id": "does-not-invent-evidence",
            "description": "Output forbids invented launch evidence.",
            "weight": 1.0,
            "failure_type": "invented_evidence",
            "passed": false,
            "missing": [
              "missing evidence"
            ],
            "present_forbidden": []
          }
        ],
        "failed": [
          {
            "id": "uses-file-backed-evidence",
            "description": "Output names file-backed source evidence.",
            "weight": 1.0,
            "failure_type": "missing_file_fixture",
            "passed": false,
            "missing": [
              "file-backed fixture",
              "input_files"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-governance",
            "description": "Output preserves governed ownership metadata.",
            "weight": 1.0,
            "failure_type": "missing_governance",
            "passed": false,
            "missing": [
              "review cadence"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-output-contract",
            "description": "Output preserves output and rollback boundaries.",
            "weight": 1.0,
            "failure_type": "missing_contract_boundary",
            "passed": false,
            "missing": [
              "output contract",
              "rollback boundary"
            ],
            "present_forbidden": []
          },
          {
            "id": "has-trust-and-scorecard",
            "description": "Output requires trust and output scorecard artifacts.",
            "weight": 1.0,
            "failure_type": "missing_release_evidence",
            "passed": false,
            "missing": [
              "trust report",
              "reports/output_quality_scorecard.md"
            ],
            "present_forbidden": []
          },
          {
            "id": "does-not-invent-evidence",
            "description": "Output forbids invented launch evidence.",
            "weight": 1.0,
            "failure_type": "invented_evidence",
            "passed": false,
            "missing": [
              "missing evidence"
            ],
            "present_forbidden": []
          }
        ]
      },
      "with_skill": {
        "score": 100.0,
        "passed_count": 5,
        "failed_count": 0,
        "checks": [
          {
            "id": "uses-file-backed-evidence",
            "description": "Output names file-backed source evidence.",
            "weight": 1.0,
            "failure_type": "missing_file_fixture",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-governance",
            "description": "Output preserves governed ownership metadata.",
            "weight": 1.0,
            "failure_type": "missing_governance",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-output-contract",
            "description": "Output preserves output and rollback boundaries.",
            "weight": 1.0,
            "failure_type": "missing_contract_boundary",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "has-trust-and-scorecard",
            "description": "Output requires trust and output scorecard artifacts.",
            "weight": 1.0,
            "failure_type": "missing_release_evidence",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          },
          {
            "id": "does-not-invent-evidence",
            "description": "Output forbids invented launch evidence.",
            "weight": 1.0,
            "failure_type": "invented_evidence",
            "passed": true,
            "missing": [],
            "present_forbidden": []
          }
        ],
        "failed": []
      },
      "delta": 100.0,
      "winner": "with_skill",
      "failure_taxonomy": []
    }
  ],
  "failures": [],
  "blind_review": {
    "pack": "reports/output_blind_review_pack.json",
    "answer_key": "reports/output_blind_answer_key.json",
    "pair_count": 5
  },
  "artifacts": {
    "json": "reports/output_quality_scorecard.json",
    "markdown": "reports/output_quality_scorecard.md",
    "blind_review_pack_json": "reports/output_blind_review_pack.json",
    "blind_review_pack_md": "reports/output_blind_review_pack.md",
    "blind_answer_key_json": "reports/output_blind_answer_key.json"
  }
}