{"aggregation":{"algorithm":"weighted-applicable-floor-v0.1","formula":"floor(sum(weight * score) / sum(applicable weight))","missingCasePolicy":"fail-closed","mixedCaseAggregation":"arithmetic-mean-basis-points"},"benchmark":{"contentIdentitySha256":"b265e14e37c869df79419f0dd74791f7fd3bc22c9f16ac50403b6df2b667d2f9","id":"GC-AGENT-TASK-BENCHMARK-v0.1","path":"benchmarks/agent_tasks/v0.1/suite.json","sha256":"9a1a6baf92a9d7450e8dbef0cb1253550d74739840c43e98c0da83d6934d1500"},"contentIdentitySha256":"f7f01dfb25cb1637e5ac3e43c4dd6df67d9b260eb11eec5e341f12e93f743a65","dimensions":[{"applicability":"all-cases","id":"semantics","metric":"exact-verification-pass-rate","weightBasisPoints":3500},{"applicability":"task-policy-step-set","id":"obligations","metric":"exact-obligation-step-pass-rate","weightBasisPoints":1200},{"applicability":"task-policy-step-set","id":"effects","metric":"reference-equal-effect-transcript","weightBasisPoints":1200},{"applicability":"all-cases","id":"patch-minimality","metric":"reference-normalized-changed-span-units","weightBasisPoints":1500},{"applicability":"all-cases","id":"resource-use","metric":"reference-normalized-deterministic-resource-units","weightBasisPoints":1300},{"applicability":"all-cases","id":"policy-scope","metric":"semantic-authority-subset","weightBasisPoints":1300}],"implementation":{"contractPath":"scripts/lib/gc_agent_scoring_contract.py","contractSha256":"140fa25fbf3b22878569a841deb52b12ed7140b60d397ab58e6bafda2b61079e","integrationTestPath":"crates/gc_cli/tests/cli_agent_benchmark_scoring.rs","integrationTestSha256":"70f8f8456d9923e9a04f7f587216f81ddbe4c7e90f41d7a58716fcff7b17c3b4","runtimePath":"scripts/lib/gc_agent_scoring.py","runtimeSha256":"62f05ec74235c9b705a140ececafc0ea147eb2867e56f6f4209d1e82aef4265b"},"kind":"genesis/agent-benchmark-scoring-v0.1","modelSpecificMetrics":{"includedInQualityScore":false,"recordedBy":"genesis/agent-benchmark-run-v0.1","separateFields":["api-cost","energy","model-latency","provider-queue-time"]},"profile":{"id":"GC-AGENT-v0.3","path":"docs/spec/GC_AGENT_PROFILE_v0.3.json","sha256":"d68dcb143067afda404e9dce3865e27b1d58e0e039a8f1a622159e467c3f2ada"},"qualityScaleBasisPoints":10000,"resourcePolicy":{"defaultEvaluatorStepLimit":200000,"maxCandidateBytes":1048576,"maxCandidateFiles":128,"maxFileBytes":262144,"maxGeneratedBytes":16777216,"maxGeneratedFiles":128,"maxStderrBytes":262144,"maxStdoutBytes":1048576,"maxVerificationSteps":8,"processTimeoutMs":30000,"resourceUnitFormula":"candidate-source-bytes + generated-artifact-bytes + normalized-canonical-json-output-or-bounded-raw-bytes + stderr-bytes + verification-command-count","wallTimeInQualityScore":false},"scoringId":"GC-AGENT-BENCHMARK-SCORING-v0.1","taskPolicies":[{"effectStepIds":[],"generatedPathPrefixes":[],"obligationStepIds":[],"policyPaths":[],"taskClass":"completion"},{"effectStepIds":[],"generatedPathPrefixes":["dist/"],"obligationStepIds":["build","package"],"policyPaths":["caps.toml"],"taskClass":"deployment"},{"effectStepIds":[],"generatedPathPrefixes":[],"obligationStepIds":[],"policyPaths":[],"taskClass":"generation"},{"effectStepIds":[],"generatedPathPrefixes":[],"obligationStepIds":["typecheck"],"policyPaths":[],"taskClass":"package-migration"},{"effectStepIds":[],"generatedPathPrefixes":[],"obligationStepIds":[],"policyPaths":[],"taskClass":"performance-repair"},{"effectStepIds":["authorized-run"],"generatedPathPrefixes":["effect.gclog"],"obligationStepIds":[],"policyPaths":["caps.toml"],"taskClass":"policy-minimization"},{"effectStepIds":[],"generatedPathPrefixes":[],"obligationStepIds":[],"policyPaths":[],"taskClass":"refactor"},{"effectStepIds":[],"generatedPathPrefixes":[],"obligationStepIds":[],"policyPaths":[],"taskClass":"repair"},{"effectStepIds":["confirm-mismatch"],"generatedPathPrefixes":[],"obligationStepIds":[],"policyPaths":["caps.toml"],"taskClass":"replay-investigation"}],"validityGate":{"editableScopeRequired":true,"failureScoreBasisPoints":0,"inapplicableDimensionPasses":true,"requiredPerfectDimensions":["semantics","obligations","policy-scope"]},"version":"0.1.0"}
