{
  "schemaVersion": 1,
  "status": "complete",
  "estimator": {
    "name": "ratioOfReducedSums",
    "trials": "Use each report definition aggregation.trials reducer within each case before summing.",
    "cases": "Sum reduced case values per arm, then compute goal-aware candidate-versus-baseline percentage.",
    "diagnostics": "Existing case-level average/min/max gains remain macro-mean diagnostics."
  },
  "benchmarks": [
    {
      "benchmarkId": "subagents",
      "sourceEvidence": "sources/provenance/subagents-cost-recalculation.json",
      "sourceSha256": "711b3265612210842bbdda9ff549a0de663d9693a8e4ef95006250b7dab600e3",
      "originalDefinition": {
        "trials": "median",
        "cases": "one case",
        "metric": "cost.total"
      },
      "publicationDefinition": {
        "trials": "median",
        "cases": "ratioOfReducedSums"
      },
      "objectives": [
        {
          "metric": "cost.total",
          "role": "primary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "subagents-codebase-exploration",
              "baselineReducedValue": 1.2856670000000001,
              "candidateReducedValue": 0.9811268400000001,
              "caseChangePercent": -23.68732805617629,
              "caseGainPercent": 23.68732805617629
            }
          ],
          "baselineReducedSum": 1.2856670000000001,
          "candidateReducedSum": 0.9811268400000001,
          "aggregateChangePercent": -23.68732805617629,
          "aggregateGainPercent": 23.68732805617629,
          "macroMeanCaseGainPercent": 23.68732805617629,
          "macroMinCaseGainPercent": 23.68732805617629,
          "macroMaxCaseGainPercent": 23.68732805617629
        }
      ]
    },
    {
      "benchmarkId": "markitdown-cost",
      "sourceReport": "benchmarks/markitdown-cost/results.json",
      "sourceSha256Before": "8199b9d2855f6145fd348189cbc849110ee52c45d92076ca51611fed52e24aac",
      "originalDefinition": {
        "revision": 5,
        "label": "MarkItDown conversion",
        "description": "Compare automatic ordinary-read conversion with the same converter installed but the extension disabled.",
        "select": {
          "suites": [
            "markitdown-cost"
          ]
        },
        "arms": {
          "baseline": "felan-no-markitdown",
          "candidate": "felan-markitdown"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          },
          {
            "metric": "usage.promptTokens",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "macroMean"
        }
      },
      "publicationDefinition": {
        "revision": 5,
        "label": "MarkItDown conversion",
        "description": "Compare automatic ordinary-read conversion with the same converter installed but the extension disabled.",
        "select": {
          "suites": [
            "markitdown-cost"
          ]
        },
        "arms": {
          "baseline": "felan-no-markitdown",
          "candidate": "felan-markitdown"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          },
          {
            "metric": "usage.promptTokens",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "ratioOfReducedSums"
        }
      },
      "objectives": [
        {
          "metric": "cost.total",
          "role": "primary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "markitdown-document-configuration",
              "baselineReducedValue": 0.020751199999999997,
              "candidateReducedValue": 0.0169152,
              "caseChangePercent": -18.485677936697634,
              "caseGainPercent": 18.485677936697634,
              "baselineAttemptTotal": 0.12115599999999999,
              "candidateAttemptTotal": 0.0550852
            },
            {
              "caseId": "markitdown-docx-extraction",
              "baselineReducedValue": 0.008628,
              "candidateReducedValue": 0.0070079999999999995,
              "caseChangePercent": -18.77607788595272,
              "caseGainPercent": 18.77607788595272,
              "baselineAttemptTotal": 0.028872,
              "candidateAttemptTotal": 0.021023999999999998
            },
            {
              "caseId": "markitdown-pptx-extraction",
              "baselineReducedValue": 0.013751999999999999,
              "candidateReducedValue": 0.007183999999999999,
              "caseChangePercent": -47.760325770796975,
              "caseGainPercent": 47.760325770796975,
              "baselineAttemptTotal": 0.041398,
              "candidateAttemptTotal": 0.021849999999999998
            },
            {
              "caseId": "markitdown-xlsx-extraction",
              "baselineReducedValue": 0.012362,
              "candidateReducedValue": 0.0071660000000000005,
              "caseChangePercent": -42.0320336515127,
              "caseGainPercent": 42.0320336515127,
              "baselineAttemptTotal": 0.035342,
              "candidateAttemptTotal": 0.021512
            }
          ],
          "baselineReducedSum": 0.05549319999999999,
          "candidateReducedSum": 0.0382732,
          "aggregateChangePercent": -31.030829002472366,
          "aggregateGainPercent": 31.030829002472366,
          "macroMeanCaseGainPercent": 31.763528811240008,
          "macroMinCaseGainPercent": 18.485677936697634,
          "macroMaxCaseGainPercent": 47.760325770796975,
          "rawAttemptTotalGainPercent": -47.31567064136033
        },
        {
          "metric": "usage.promptTokens",
          "role": "secondary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "markitdown-document-configuration",
              "baselineReducedValue": 9256,
              "candidateReducedValue": 9070,
              "caseChangePercent": -2.0095073465859983,
              "caseGainPercent": 2.0095073465859983,
              "baselineAttemptTotal": 134996,
              "candidateAttemptTotal": 27923
            },
            {
              "caseId": "markitdown-docx-extraction",
              "baselineReducedValue": 3000,
              "candidateReducedValue": 3192,
              "caseChangePercent": 6.4,
              "caseGainPercent": -6.4,
              "baselineAttemptTotal": 10476,
              "candidateAttemptTotal": 9576
            },
            {
              "caseId": "markitdown-pptx-extraction",
              "baselineReducedValue": 4710,
              "candidateReducedValue": 3172,
              "caseChangePercent": -32.65392781316348,
              "caseGainPercent": 32.65392781316348,
              "baselineAttemptTotal": 14147,
              "candidateAttemptTotal": 9539
            },
            {
              "caseId": "markitdown-xlsx-extraction",
              "baselineReducedValue": 4603,
              "candidateReducedValue": 3169,
              "caseChangePercent": -31.15359548120791,
              "caseGainPercent": 31.15359548120791,
              "baselineAttemptTotal": 12427,
              "candidateAttemptTotal": 9508
            }
          ],
          "baselineReducedSum": 21569,
          "candidateReducedSum": 18603,
          "aggregateChangePercent": -13.751217024433215,
          "aggregateGainPercent": 13.751217024433215,
          "macroMeanCaseGainPercent": 14.854257660239346,
          "macroMinCaseGainPercent": -6.4,
          "macroMaxCaseGainPercent": 32.65392781316348,
          "rawAttemptTotalGainPercent": -67.13320856050126
        }
      ],
      "publishedReportSha256": "d1ac1289ba94b1122bb98d69a1806bfd8d8a7478a983ca48bb51db01d92175e2"
    },
    {
      "benchmarkId": "output-style-concise",
      "sourceReport": "benchmarks/output-style-concise/results.json",
      "sourceSha256Before": "0f44fb5f500ede4a5cc92abd4bfae0a130b123be4e3ba839399ce7ce3a1aaad3",
      "originalDefinition": {
        "revision": 7,
        "label": "Concise output style",
        "description": "Compare concise output with output style disabled while preserving required information.",
        "select": {
          "cases": [
            "output-style-planning",
            "output-style-review",
            "output-style-support"
          ]
        },
        "arms": {
          "baseline": "felan-no-output-style",
          "candidate": "felan-concise"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          },
          {
            "metric": "usage.outputTokens",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "macroMean"
        }
      },
      "publicationDefinition": {
        "revision": 7,
        "label": "Concise output style",
        "description": "Compare concise output with output style disabled while preserving required information.",
        "select": {
          "cases": [
            "output-style-planning",
            "output-style-review",
            "output-style-support"
          ]
        },
        "arms": {
          "baseline": "felan-no-output-style",
          "candidate": "felan-concise"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          },
          {
            "metric": "usage.outputTokens",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "ratioOfReducedSums"
        }
      },
      "objectives": [
        {
          "metric": "cost.total",
          "role": "primary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "output-style-planning",
              "baselineReducedValue": 0.080427,
              "candidateReducedValue": 0.065847,
              "caseChangePercent": -18.12824051624454,
              "caseGainPercent": 18.12824051624454,
              "baselineAttemptTotal": 0.265072,
              "candidateAttemptTotal": 0.22307400000000002
            },
            {
              "caseId": "output-style-review",
              "baselineReducedValue": 0.06596400000000001,
              "candidateReducedValue": 0.053063,
              "caseChangePercent": -19.557637499242023,
              "caseGainPercent": 19.557637499242023,
              "baselineAttemptTotal": 0.192224,
              "candidateAttemptTotal": 0.19565600000000002
            },
            {
              "caseId": "output-style-support",
              "baselineReducedValue": 0.08315700000000001,
              "candidateReducedValue": 0.077348,
              "caseChangePercent": -6.985581490433768,
              "caseGainPercent": 6.985581490433768,
              "baselineAttemptTotal": 0.242291,
              "candidateAttemptTotal": 0.21230500000000002
            }
          ],
          "baselineReducedSum": 0.229548,
          "candidateReducedSum": 0.196258,
          "aggregateChangePercent": -14.502413438583659,
          "aggregateGainPercent": 14.502413438583659,
          "macroMeanCaseGainPercent": 14.890486501973443,
          "macroMinCaseGainPercent": 6.985581490433768,
          "macroMaxCaseGainPercent": 19.557637499242023,
          "rawAttemptTotalGainPercent": -9.798924222434085
        },
        {
          "metric": "usage.outputTokens",
          "role": "secondary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "output-style-planning",
              "baselineReducedValue": 1626,
              "candidateReducedValue": 1388,
              "caseChangePercent": -14.637146371463713,
              "caseGainPercent": 14.637146371463713,
              "baselineAttemptTotal": 4916,
              "candidateAttemptTotal": 4617
            },
            {
              "caseId": "output-style-review",
              "baselineReducedValue": 1034,
              "candidateReducedValue": 909,
              "caseChangePercent": -12.088974854932303,
              "caseGainPercent": 12.088974854932303,
              "baselineAttemptTotal": 3463,
              "candidateAttemptTotal": 3229
            },
            {
              "caseId": "output-style-support",
              "baselineReducedValue": 1527,
              "candidateReducedValue": 1203,
              "caseChangePercent": -21.218074656188605,
              "caseGainPercent": 21.218074656188605,
              "baselineAttemptTotal": 4504,
              "candidateAttemptTotal": 3858
            }
          ],
          "baselineReducedSum": 4187,
          "candidateReducedSum": 3500,
          "aggregateChangePercent": -16.40792930499164,
          "aggregateGainPercent": 16.40792930499164,
          "macroMeanCaseGainPercent": 15.981398627528208,
          "macroMinCaseGainPercent": 12.088974854932303,
          "macroMaxCaseGainPercent": 21.218074656188605,
          "rawAttemptTotalGainPercent": -9.151595125359
        }
      ],
      "publishedReportSha256": "247fada4ab4008dda78b73223c7f1363be2f9a2e7ec05445d58867fa9d825eac"
    },
    {
      "benchmarkId": "prewalk",
      "sourceReport": "benchmarks/prewalk/results.json",
      "sourceSha256Before": "c5abca0c68fa3071d0205df8f8af9860b12087a3ff81401b4a6ce5470c26d99c",
      "originalDefinition": {
        "revision": 11,
        "label": "Prewalk organic routing cost",
        "description": "Compare organic Prewalk routing with Prewalk disabled; candidate attempts without a successful entry are invalid.",
        "select": {
          "cases": [
            "prewalk-checkout",
            "memory-summary-links"
          ]
        },
        "arms": {
          "baseline": "felan-no-prewalk",
          "candidate": "felan-prewalk"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "macroMean"
        }
      },
      "publicationDefinition": {
        "revision": 11,
        "label": "Prewalk organic routing cost",
        "description": "Compare organic Prewalk routing with Prewalk disabled; candidate attempts without a successful entry are invalid.",
        "select": {
          "cases": [
            "prewalk-checkout",
            "memory-summary-links"
          ]
        },
        "arms": {
          "baseline": "felan-no-prewalk",
          "candidate": "felan-prewalk"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "ratioOfReducedSums"
        }
      },
      "objectives": [
        {
          "metric": "cost.total",
          "role": "primary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "memory-summary-links",
              "baselineReducedValue": 6.814177999999999,
              "candidateReducedValue": 1.93362268,
              "caseChangePercent": -71.62353727771713,
              "caseGainPercent": 71.62353727771713,
              "baselineAttemptTotal": 20.439731000000002,
              "candidateAttemptTotal": 5.53533596
            },
            {
              "caseId": "prewalk-checkout",
              "baselineReducedValue": 1.710072,
              "candidateReducedValue": 0.9660633200000002,
              "caseChangePercent": -43.50744763963154,
              "caseGainPercent": 43.50744763963154,
              "baselineAttemptTotal": 5.112296000000001,
              "candidateAttemptTotal": 2.77146096
            }
          ],
          "baselineReducedSum": 8.524249999999999,
          "candidateReducedSum": 2.899686,
          "aggregateChangePercent": -65.98309528697538,
          "aggregateGainPercent": 65.98309528697538,
          "macroMeanCaseGainPercent": 57.56549245867434,
          "macroMinCaseGainPercent": 43.50744763963154,
          "macroMaxCaseGainPercent": 71.62353727771713,
          "rawAttemptTotalGainPercent": -67.49065379431542
        }
      ],
      "publishedReportSha256": "27d856ac470ad6865892405dac695721761a86cc5b396c061789ab60b55c2951"
    },
    {
      "benchmarkId": "rtk",
      "sourceReport": "benchmarks/rtk/results.json",
      "sourceSha256Before": "760b96cc2e24e59acff3c7bef94deb05d651033c4f991e51cc56aed01c0b88f7",
      "originalDefinition": {
        "revision": 12,
        "label": "RTK context reduction",
        "select": {
          "cases": [
            "rtk-felan-project-instructions",
            "prewalk-checkout"
          ]
        },
        "arms": {
          "baseline": "felan-no-rtk-optimizer",
          "candidate": "felan-rtk-optimizer"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          },
          {
            "metric": "usage.promptTokens",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "macroMean"
        }
      },
      "publicationDefinition": {
        "revision": 12,
        "label": "RTK context reduction",
        "select": {
          "cases": [
            "rtk-felan-project-instructions",
            "prewalk-checkout"
          ]
        },
        "arms": {
          "baseline": "felan-no-rtk-optimizer",
          "candidate": "felan-rtk-optimizer"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          },
          {
            "metric": "usage.promptTokens",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "ratioOfReducedSums"
        }
      },
      "objectives": [
        {
          "metric": "cost.total",
          "role": "primary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "prewalk-checkout",
              "baselineReducedValue": 2.2536910000000003,
              "candidateReducedValue": 1.906687,
              "caseChangePercent": -15.397141844201368,
              "caseGainPercent": 15.397141844201368,
              "baselineAttemptTotal": 6.977328999999999,
              "candidateAttemptTotal": 6.1940539999999995
            },
            {
              "caseId": "rtk-felan-project-instructions",
              "baselineReducedValue": 4.006117999999999,
              "candidateReducedValue": 2.6902200000000005,
              "caseChangePercent": -32.84721019201129,
              "caseGainPercent": 32.84721019201129,
              "baselineAttemptTotal": 12.60722,
              "candidateAttemptTotal": 8.121762
            }
          ],
          "baselineReducedSum": 6.259808999999999,
          "candidateReducedSum": 4.596907000000001,
          "aggregateChangePercent": -26.564740234086987,
          "aggregateGainPercent": 26.564740234086987,
          "macroMeanCaseGainPercent": 24.12217601810633,
          "macroMinCaseGainPercent": 15.397141844201368,
          "macroMaxCaseGainPercent": 32.84721019201129,
          "rawAttemptTotalGainPercent": -26.902498495114692
        },
        {
          "metric": "usage.promptTokens",
          "role": "secondary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "prewalk-checkout",
              "baselineReducedValue": 1917065,
              "candidateReducedValue": 1507859,
              "caseChangePercent": -21.345442121159167,
              "caseGainPercent": 21.345442121159167,
              "baselineAttemptTotal": 6366377,
              "candidateAttemptTotal": 4453168
            },
            {
              "caseId": "rtk-felan-project-instructions",
              "baselineReducedValue": 5323378,
              "candidateReducedValue": 2794248,
              "caseChangePercent": -47.50987061223156,
              "caseGainPercent": 47.50987061223156,
              "baselineAttemptTotal": 16572160,
              "candidateAttemptTotal": 8251398
            }
          ],
          "baselineReducedSum": 7240443,
          "candidateReducedSum": 4302107,
          "aggregateChangePercent": -40.582268239664344,
          "aggregateGainPercent": 40.582268239664344,
          "macroMeanCaseGainPercent": 34.42765636669536,
          "macroMinCaseGainPercent": 21.345442121159167,
          "macroMaxCaseGainPercent": 47.50987061223156,
          "rawAttemptTotalGainPercent": -44.614750278101866
        }
      ],
      "publishedReportSha256": "607907919f032cb05553cd07abc282349064ea74f247b59fa61c9b358917a9ab"
    },
    {
      "benchmarkId": "codebase-memory",
      "sourceEvidence": "sources/provenance/codebase-memory.json",
      "sourceSha256": "d712f29701196491246c803766c8f3a482e64236b4e3798caf0d235cceb1ac11",
      "originalDefinition": {
        "trials": "median",
        "cases": "three cases from three --case batches",
        "metrics": [
          "cost.total",
          "duration.stepsMs"
        ]
      },
      "publicationDefinition": {
        "trials": "median",
        "cases": "ratioOfReducedSums"
      },
      "objectives": [
        {
          "metric": "cost.total",
          "role": "primary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "cbm-extension-architecture",
              "baselineReducedValue": 1.2001530000000002,
              "candidateReducedValue": 1.2125380000000001,
              "caseChangePercent": 1.031951759483988,
              "caseGainPercent": -1.031951759483988
            },
            {
              "caseId": "cbm-extension-config-scope",
              "baselineReducedValue": 3.8947869999999996,
              "candidateReducedValue": 2.243165,
              "caseChangePercent": -42.405964690752015,
              "caseGainPercent": 42.405964690752015
            },
            {
              "caseId": "cbm-project-instructions",
              "baselineReducedValue": 0.632202,
              "candidateReducedValue": 0.680038,
              "caseChangePercent": 7.566568913100557,
              "caseGainPercent": -7.566568913100557
            }
          ],
          "baselineReducedSum": 5.727142,
          "candidateReducedSum": 4.135741,
          "aggregateChangePercent": -27.78700091598915,
          "aggregateGainPercent": 27.78700091598915,
          "macroMeanCaseGainPercent": 11.269148006055824,
          "macroMinCaseGainPercent": -7.566568913100557,
          "macroMaxCaseGainPercent": 42.405964690752015
        },
        {
          "metric": "duration.stepsMs",
          "role": "secondary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "cbm-extension-architecture",
              "baselineReducedValue": 233785,
              "candidateReducedValue": 244129,
              "caseChangePercent": 4.424578138032808,
              "caseGainPercent": -4.424578138032808
            },
            {
              "caseId": "cbm-extension-config-scope",
              "baselineReducedValue": 727694,
              "candidateReducedValue": 666171,
              "caseChangePercent": -8.454515222057623,
              "caseGainPercent": 8.454515222057623
            },
            {
              "caseId": "cbm-project-instructions",
              "baselineReducedValue": 205803,
              "candidateReducedValue": 166419,
              "caseChangePercent": -19.136747277736475,
              "caseGainPercent": 19.136747277736475
            }
          ],
          "baselineReducedSum": 1167282,
          "candidateReducedSum": 1076719,
          "aggregateChangePercent": -7.758450828505879,
          "aggregateGainPercent": 7.758450828505879,
          "macroMeanCaseGainPercent": 7.722228120587097,
          "macroMinCaseGainPercent": -4.424578138032808,
          "macroMaxCaseGainPercent": 19.136747277736475
        }
      ]
    },
    {
      "benchmarkId": "session-compaction",
      "sourceReport": "benchmarks/session-compaction/results.json",
      "sourceSha256Before": "ea99f9458eb368e9e8731506c43fda2925236527c5ff4eef0e8afbefeca292e9",
      "originalDefinition": {
        "revision": 4,
        "label": "Inherited-model session compaction",
        "description": "Compare native Pi compaction with Felan's verified summary using the default inherited active model on a sanitized real coding trajectory.",
        "select": {
          "suites": [
            "session-compaction"
          ]
        },
        "arms": {
          "baseline": "felan-no-session-compaction",
          "candidate": "felan-inherit-compaction"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          },
          {
            "metric": "usage.promptTokens",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "ratioOfReducedSums"
        }
      },
      "publicationDefinition": {
        "revision": 4,
        "label": "Inherited-model session compaction",
        "description": "Compare native Pi compaction with Felan's verified summary using the default inherited active model on a sanitized real coding trajectory.",
        "select": {
          "suites": [
            "session-compaction"
          ]
        },
        "arms": {
          "baseline": "felan-no-session-compaction",
          "candidate": "felan-inherit-compaction"
        },
        "trials": 3,
        "qualityGates": [
          {
            "metric": "quality.passRate",
            "min": 0.8
          }
        ],
        "objective": [
          {
            "metric": "cost.total",
            "goal": "minimize"
          },
          {
            "metric": "usage.promptTokens",
            "goal": "minimize"
          }
        ],
        "aggregation": {
          "trials": "median",
          "cases": "ratioOfReducedSums"
        }
      },
      "objectives": [
        {
          "metric": "cost.total",
          "role": "primary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "session-compaction-continuation",
              "baselineReducedValue": 0.49551,
              "candidateReducedValue": 0.27131000000000005,
              "caseChangePercent": -45.24631188068857,
              "caseGainPercent": 45.24631188068857
            }
          ],
          "baselineReducedSum": 0.49551,
          "candidateReducedSum": 0.27131000000000005,
          "aggregateChangePercent": -45.24631188068857,
          "aggregateGainPercent": 45.24631188068857,
          "macroMeanCaseGainPercent": 45.24631188068857,
          "macroMinCaseGainPercent": 45.24631188068857,
          "macroMaxCaseGainPercent": 45.24631188068857
        },
        {
          "metric": "usage.promptTokens",
          "role": "secondary",
          "goal": "minimize",
          "trialReducer": "median",
          "cases": [
            {
              "caseId": "session-compaction-continuation",
              "baselineReducedValue": 83304,
              "candidateReducedValue": 39891,
              "caseChangePercent": -52.11394410832612,
              "caseGainPercent": 52.11394410832612
            }
          ],
          "baselineReducedSum": 83304,
          "candidateReducedSum": 39891,
          "aggregateChangePercent": -52.11394410832612,
          "aggregateGainPercent": 52.11394410832612,
          "macroMeanCaseGainPercent": 52.11394410832612,
          "macroMinCaseGainPercent": 52.11394410832612,
          "macroMaxCaseGainPercent": 52.11394410832612
        }
      ],
      "quality": {
        "baseline": {
          "attemptCount": 3,
          "passed": 2,
          "failed": 1,
          "passRate": 0.6666666666666666
        },
        "candidate": {
          "attemptCount": 3,
          "passed": 3,
          "failed": 0,
          "passRate": 1.0
        },
        "candidateGatePassed": true,
        "baselineDisclosure": "One native continuation omitted the required black-box end-to-end publication plan; the candidate quality gate is evaluated independently and passed 3/3."
      },
      "publicationNote": "The candidate quality gate passed 3/3. The metric is included in the descriptive portfolio total; the native baseline 2/3 result and local pre-release provenance remain disclosed."
    }
  ]
}