> ## Documentation Index > Fetch the complete documentation index at: https://evalgate.com/docs/llms.txt > Use this file to discover all available pages before exploring further. # List quality quality > Requires scopes: runs:read ## OpenAPI ````yaml /api-reference/openapi.json get /api/quality openapi: 3.1.0 info: title: EvalGate API version: 3.7.6 description: EvalGate API. See docs/api-contract.md for stability commitments. servers: - url: https://evalgate.com description: Production - url: http://localhost:3000 description: Local security: [] paths: /api/quality: get: tags: - quality summary: List quality quality description: 'Requires scopes: runs:read' operationId: get_quality parameters: - name: limit in: query required: false schema: type: integer minimum: 1 maximum: 1000 default: 50 - name: offset in: query required: false schema: type: integer minimum: 0 default: 0 responses: '200': description: Successful response content: application/json: schema: $ref: '#/components/schemas/GetApiQualityResponse200ApplicationJson' '400': $ref: '#/components/responses/ValidationError' '401': $ref: '#/components/responses/Unauthorized' '403': $ref: '#/components/responses/Forbidden' '404': $ref: '#/components/responses/NotFound' '409': $ref: '#/components/responses/Conflict' '429': $ref: '#/components/responses/RateLimited' '500': $ref: '#/components/responses/InternalError' security: - bearerAuth: - runs:read components: schemas: GetApiQualityResponse200ApplicationJson: oneOf: - type: object properties: evaluationId: type: number evaluationRunId: oneOf: - type: 'null' - type: number integrity: oneOf: - type: 'null' - type: object properties: summary: type: object properties: byBehavior: type: object properties: tool_use: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings classification: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings retrieval: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings extraction: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings multi_step_reasoning: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - tool_use - classification - retrieval - extraction - multi_step_reasoning byDifficulty: type: object properties: medium: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings hard: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings easy: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings flaky: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - medium - hard - easy - flaky trajectory: type: object properties: efficiency: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings toolUsage: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - efficiency - toolUsage workflow: type: object properties: axes: type: object properties: autonomyFit: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings escalationQuality: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings approvalSafety: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings hookCompliance: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings memoryCorrectness: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings recoveryQuality: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings handoffQuality: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings debuggability: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - autonomyFit - escalationQuality - approvalSafety - hookCompliance - memoryCorrectness - recoveryQuality - handoffQuality - debuggability slices: type: object properties: byAutonomy: type: object properties: manual: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings default_approvals: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings bypass_approvals: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings autopilot: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - manual - default_approvals - bypass_approvals - autopilot byWorkflowClass: type: object properties: approval_gated_mutation: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings background_execution: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings recovery_required: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings policy_conflict: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings handoff_heavy: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings memory_sensitive: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings parallel_tool_use: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - approval_gated_mutation - background_execution - recovery_required - policy_conflict - handoff_heavy - memory_sensitive - parallel_tool_use additionalProperties: false required: - byAutonomy - byWorkflowClass additionalProperties: false required: - axes - slices metadata: type: object properties: totalCases: type: number evaluatedAt: type: string evaluationRunId: oneOf: - type: 'null' - type: number uncategorizedBehaviorCount: type: number uncategorizedDifficultyCount: type: number additionalProperties: false required: - totalCases - evaluatedAt - evaluationRunId - uncategorizedBehaviorCount - uncategorizedDifficultyCount additionalProperties: false required: - byBehavior - byDifficulty - metadata datasetIntegrity: type: object properties: behaviorDistribution: type: object properties: tool_use: type: number classification: type: number retrieval: type: number extraction: type: number multi_step_reasoning: type: number additionalProperties: false required: - tool_use - classification - retrieval - extraction - multi_step_reasoning difficultyDistribution: type: object properties: medium: type: number hard: type: number easy: type: number flaky: type: number additionalProperties: false required: - medium - hard - easy - flaky failureOriginShare: type: number syntheticOnly: type: boolean coverageWarnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - behaviorDistribution - difficultyDistribution - failureOriginShare - syntheticOnly - coverageWarnings metricValidity: type: object properties: historicalSnapshotsChecked: type: number driftSignals: type: array items: type: object properties: metric: type: string currentMean: type: number previousMean: type: number delta: type: number triggered: type: boolean additionalProperties: false required: - metric - currentMean - previousMean - delta - triggered improvementWithoutFailureReduction: type: boolean additionalProperties: false required: - historicalSnapshotsChecked - driftSignals - improvementWithoutFailureReduction antiGaming: type: object properties: improvedEasyBucketOnly: type: boolean failureDistributionStagnant: type: boolean redundancyIncreaseWithScoreGain: type: boolean detected: type: boolean additionalProperties: false required: - improvedEasyBucketOnly - failureDistributionStagnant - redundancyIncreaseWithScoreGain - detected coverageIntegrity: type: object properties: coverageByBehavior: type: object properties: tool_use: type: number classification: type: number retrieval: type: number extraction: type: number multi_step_reasoning: type: number additionalProperties: false required: - tool_use - classification - retrieval - extraction - multi_step_reasoning dominantFailureMode: type: object properties: mode: oneOf: - type: 'null' - type: string share: type: number additionalProperties: false required: - mode - share lowFailureDiversity: type: boolean additionalProperties: false required: - coverageByBehavior - dominantFailureMode - lowFailureDiversity lifecycle: type: object properties: counts: type: object properties: active: type: number validated: type: number superseded: type: number deprecated: type: number archived: type: number candidate: type: number active_gate: type: number additionalProperties: false required: - active - validated - superseded - deprecated - archived - candidate - active_gate recommendedTransitions: type: array items: type: object properties: testCaseId: type: number currentState: oneOf: - type: string enum: - active - type: string enum: - validated - type: string enum: - superseded - type: string enum: - deprecated - type: string enum: - archived - type: string enum: - candidate - type: string enum: - active_gate nextState: oneOf: - type: string enum: - active - type: string enum: - validated - type: string enum: - superseded - type: string enum: - deprecated - type: string enum: - archived - type: string enum: - candidate - type: string enum: - active_gate reason: oneOf: - type: string enum: - manual_approval - type: string enum: - consistent_failures - type: string enum: - activated_for_ci - type: string enum: - low_signal - type: string enum: - unused_for_cycles - type: string enum: - reactivated_by_failure - type: string enum: - reactivated_by_manual_review - type: string enum: - no_transition additionalProperties: false required: - testCaseId - currentState - nextState - reason additionalProperties: false required: - counts - recommendedTransitions warnings: type: array items: type: object properties: code: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch severity: oneOf: - type: string enum: - critical - type: string enum: - info - type: string enum: - warning message: type: string context: type: object additionalProperties: $ref: '#/components/schemas/JsonValue' additionalProperties: false required: - code - severity - message events: type: array items: type: object properties: type: oneOf: - type: string enum: - metric_drift_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_integrity_warning - type: string enum: - eval_case_promoted - type: string enum: - eval_case_deprecated - type: string enum: - eval_case_archived - type: string enum: - eval_case_reactivated message: type: string context: type: object additionalProperties: $ref: '#/components/schemas/JsonValue' additionalProperties: false required: - type - message metadata: type: object properties: totalWarnings: type: number refreshedAt: type: string sourceTrajectoryMetricsAvailable: type: boolean overallPassRate: type: number averageTrajectoryRedundancy: oneOf: - type: 'null' - type: number additionalProperties: false required: - totalWarnings - refreshedAt - sourceTrajectoryMetricsAvailable - overallPassRate - averageTrajectoryRedundancy additionalProperties: false required: - summary - datasetIntegrity - metricValidity - antiGaming - coverageIntegrity - lifecycle - warnings - events - metadata workflowCoverage: oneOf: - type: 'null' - type: object properties: totalCases: type: number agentWorkflowCases: type: number coverageRate: type: number workflowClassDistribution: type: object properties: approval_gated_mutation: type: number background_execution: type: number recovery_required: type: number policy_conflict: type: number handoff_heavy: type: number memory_sensitive: type: number parallel_tool_use: type: number additionalProperties: false required: - approval_gated_mutation - background_execution - recovery_required - policy_conflict - handoff_heavy - memory_sensitive - parallel_tool_use autonomyDistribution: type: object properties: manual: type: number default_approvals: type: number bypass_approvals: type: number autopilot: type: number additionalProperties: false required: - manual - default_approvals - bypass_approvals - autopilot generatedCandidateCount: type: number experimentRecipeCount: type: number gaps: type: array items: type: object properties: id: type: string workflowClass: oneOf: - type: string enum: - approval_gated_mutation - type: string enum: - background_execution - type: string enum: - recovery_required - type: string enum: - policy_conflict - type: string enum: - handoff_heavy - type: string enum: - memory_sensitive - type: string enum: - parallel_tool_use severity: oneOf: - type: string enum: - critical - type: string enum: - info - type: string enum: - warning description: type: string recommendation: type: string additionalProperties: false required: - id - workflowClass - severity - description - recommendation additionalProperties: false required: - totalCases - agentWorkflowCases - coverageRate - workflowClassDistribution - autonomyDistribution - generatedCandidateCount - experimentRecipeCount - gaps qualityScore: oneOf: - type: 'null' - type: object properties: score: type: number breakdown: $ref: '#/components/schemas/JsonValue' flags: $ref: '#/components/schemas/JsonValue' evidenceLevel: oneOf: - type: 'null' - type: string scoringVersion: type: string traceCoverageRate: oneOf: - type: 'null' - type: string provenanceCoverageRate: oneOf: - type: 'null' - type: string model: oneOf: - type: 'null' - type: string baselineScore: oneOf: - type: 'null' - type: number regressionDelta: oneOf: - type: 'null' - type: number regressionDetected: type: boolean baselineMissing: type: boolean avgLatencyMs: oneOf: - type: 'null' - $ref: '#/components/schemas/JsonValue' costUsd: oneOf: - type: 'null' - $ref: '#/components/schemas/JsonValue' baselineCostUsd: oneOf: - type: 'null' - $ref: '#/components/schemas/JsonValue' createdAt: type: string format: date-time additionalProperties: false required: - score - breakdown - flags - evidenceLevel - scoringVersion - traceCoverageRate - provenanceCoverageRate - model - baselineScore - regressionDelta - regressionDetected - baselineMissing - avgLatencyMs - costUsd - baselineCostUsd - createdAt message: oneOf: - type: 'null' - type: string legacy: type: object additionalProperties: $ref: '#/components/schemas/JsonValue' additionalProperties: false required: - evaluationId - evaluationRunId - integrity - workflowCoverage - qualityScore - message - legacy - type: object properties: evaluationId: type: number integrity: oneOf: - type: 'null' - type: object properties: summary: type: object properties: byBehavior: type: object properties: tool_use: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings classification: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings retrieval: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings extraction: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings multi_step_reasoning: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - tool_use - classification - retrieval - extraction - multi_step_reasoning byDifficulty: type: object properties: medium: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings hard: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings easy: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings flaky: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - medium - hard - easy - flaky trajectory: type: object properties: efficiency: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings toolUsage: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - efficiency - toolUsage workflow: type: object properties: axes: type: object properties: autonomyFit: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings escalationQuality: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings approvalSafety: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings hookCompliance: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings memoryCorrectness: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings recoveryQuality: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings handoffQuality: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings debuggability: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - autonomyFit - escalationQuality - approvalSafety - hookCompliance - memoryCorrectness - recoveryQuality - handoffQuality - debuggability slices: type: object properties: byAutonomy: type: object properties: manual: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings default_approvals: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings bypass_approvals: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings autopilot: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - manual - default_approvals - bypass_approvals - autopilot byWorkflowClass: type: object properties: approval_gated_mutation: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings background_execution: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings recovery_required: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings policy_conflict: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings handoff_heavy: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings memory_sensitive: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings parallel_tool_use: type: object properties: mean: type: number stdDev: type: number 'n': type: number confidenceInterval: oneOf: - type: 'null' - type: object properties: lower: type: number upper: type: number method: oneOf: - type: string enum: - normal - type: string enum: - wilson additionalProperties: false required: - lower - upper - method composition: type: object properties: percentageOfDataset: type: number additionalProperties: false required: - percentageOfDataset status: oneOf: - type: string enum: - valid - type: string enum: - invalid - type: string enum: - unstable warnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - mean - stdDev - 'n' - confidenceInterval - composition - status - warnings additionalProperties: false required: - approval_gated_mutation - background_execution - recovery_required - policy_conflict - handoff_heavy - memory_sensitive - parallel_tool_use additionalProperties: false required: - byAutonomy - byWorkflowClass additionalProperties: false required: - axes - slices metadata: type: object properties: totalCases: type: number evaluatedAt: type: string evaluationRunId: oneOf: - type: 'null' - type: number uncategorizedBehaviorCount: type: number uncategorizedDifficultyCount: type: number additionalProperties: false required: - totalCases - evaluatedAt - evaluationRunId - uncategorizedBehaviorCount - uncategorizedDifficultyCount additionalProperties: false required: - byBehavior - byDifficulty - metadata datasetIntegrity: type: object properties: behaviorDistribution: type: object properties: tool_use: type: number classification: type: number retrieval: type: number extraction: type: number multi_step_reasoning: type: number additionalProperties: false required: - tool_use - classification - retrieval - extraction - multi_step_reasoning difficultyDistribution: type: object properties: medium: type: number hard: type: number easy: type: number flaky: type: number additionalProperties: false required: - medium - hard - easy - flaky failureOriginShare: type: number syntheticOnly: type: boolean coverageWarnings: type: array items: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch additionalProperties: false required: - behaviorDistribution - difficultyDistribution - failureOriginShare - syntheticOnly - coverageWarnings metricValidity: type: object properties: historicalSnapshotsChecked: type: number driftSignals: type: array items: type: object properties: metric: type: string currentMean: type: number previousMean: type: number delta: type: number triggered: type: boolean additionalProperties: false required: - metric - currentMean - previousMean - delta - triggered improvementWithoutFailureReduction: type: boolean additionalProperties: false required: - historicalSnapshotsChecked - driftSignals - improvementWithoutFailureReduction antiGaming: type: object properties: improvedEasyBucketOnly: type: boolean failureDistributionStagnant: type: boolean redundancyIncreaseWithScoreGain: type: boolean detected: type: boolean additionalProperties: false required: - improvedEasyBucketOnly - failureDistributionStagnant - redundancyIncreaseWithScoreGain - detected coverageIntegrity: type: object properties: coverageByBehavior: type: object properties: tool_use: type: number classification: type: number retrieval: type: number extraction: type: number multi_step_reasoning: type: number additionalProperties: false required: - tool_use - classification - retrieval - extraction - multi_step_reasoning dominantFailureMode: type: object properties: mode: oneOf: - type: 'null' - type: string share: type: number additionalProperties: false required: - mode - share lowFailureDiversity: type: boolean additionalProperties: false required: - coverageByBehavior - dominantFailureMode - lowFailureDiversity lifecycle: type: object properties: counts: type: object properties: active: type: number validated: type: number superseded: type: number deprecated: type: number archived: type: number candidate: type: number active_gate: type: number additionalProperties: false required: - active - validated - superseded - deprecated - archived - candidate - active_gate recommendedTransitions: type: array items: type: object properties: testCaseId: type: number currentState: oneOf: - type: string enum: - active - type: string enum: - validated - type: string enum: - superseded - type: string enum: - deprecated - type: string enum: - archived - type: string enum: - candidate - type: string enum: - active_gate nextState: oneOf: - type: string enum: - active - type: string enum: - validated - type: string enum: - superseded - type: string enum: - deprecated - type: string enum: - archived - type: string enum: - candidate - type: string enum: - active_gate reason: oneOf: - type: string enum: - manual_approval - type: string enum: - consistent_failures - type: string enum: - activated_for_ci - type: string enum: - low_signal - type: string enum: - unused_for_cycles - type: string enum: - reactivated_by_failure - type: string enum: - reactivated_by_manual_review - type: string enum: - no_transition additionalProperties: false required: - testCaseId - currentState - nextState - reason additionalProperties: false required: - counts - recommendedTransitions warnings: type: array items: type: object properties: code: oneOf: - type: string enum: - metric_missing_data - type: string enum: - metric_low_confidence - type: string enum: - metric_invalid_sample - type: string enum: - metric_high_variance - type: string enum: - metric_drift_detected - type: string enum: - metric_inconsistency_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_behavior_imbalance - type: string enum: - dataset_difficulty_imbalance - type: string enum: - dataset_behavior_unstable - type: string enum: - dataset_failure_origin_low - type: string enum: - dataset_synthetic_only - type: string enum: - coverage_gap_detected - type: string enum: - coverage_cluster_dominance - type: string enum: - shadow_harness_validation_failed - type: string enum: - shadow_harness_contract_mismatch severity: oneOf: - type: string enum: - critical - type: string enum: - info - type: string enum: - warning message: type: string context: type: object additionalProperties: $ref: '#/components/schemas/JsonValue' additionalProperties: false required: - code - severity - message events: type: array items: type: object properties: type: oneOf: - type: string enum: - metric_drift_detected - type: string enum: - metric_gaming_detected - type: string enum: - dataset_integrity_warning - type: string enum: - eval_case_promoted - type: string enum: - eval_case_deprecated - type: string enum: - eval_case_archived - type: string enum: - eval_case_reactivated message: type: string context: type: object additionalProperties: $ref: '#/components/schemas/JsonValue' additionalProperties: false required: - type - message metadata: type: object properties: totalWarnings: type: number refreshedAt: type: string sourceTrajectoryMetricsAvailable: type: boolean overallPassRate: type: number averageTrajectoryRedundancy: oneOf: - type: 'null' - type: number additionalProperties: false required: - totalWarnings - refreshedAt - sourceTrajectoryMetricsAvailable - overallPassRate - averageTrajectoryRedundancy additionalProperties: false required: - summary - datasetIntegrity - metricValidity - antiGaming - coverageIntegrity - lifecycle - warnings - events - metadata workflowCoverage: oneOf: - type: 'null' - type: object properties: totalCases: type: number agentWorkflowCases: type: number coverageRate: type: number workflowClassDistribution: type: object properties: approval_gated_mutation: type: number background_execution: type: number recovery_required: type: number policy_conflict: type: number handoff_heavy: type: number memory_sensitive: type: number parallel_tool_use: type: number additionalProperties: false required: - approval_gated_mutation - background_execution - recovery_required - policy_conflict - handoff_heavy - memory_sensitive - parallel_tool_use autonomyDistribution: type: object properties: manual: type: number default_approvals: type: number bypass_approvals: type: number autopilot: type: number additionalProperties: false required: - manual - default_approvals - bypass_approvals - autopilot generatedCandidateCount: type: number experimentRecipeCount: type: number gaps: type: array items: type: object properties: id: type: string workflowClass: oneOf: - type: string enum: - approval_gated_mutation - type: string enum: - background_execution - type: string enum: - recovery_required - type: string enum: - policy_conflict - type: string enum: - handoff_heavy - type: string enum: - memory_sensitive - type: string enum: - parallel_tool_use severity: oneOf: - type: string enum: - critical - type: string enum: - info - type: string enum: - warning description: type: string recommendation: type: string additionalProperties: false required: - id - workflowClass - severity - description - recommendation additionalProperties: false required: - totalCases - agentWorkflowCases - coverageRate - workflowClassDistribution - autonomyDistribution - generatedCandidateCount - experimentRecipeCount - gaps qualityScoreTrend: type: array items: type: object properties: id: type: number evaluationRunId: type: number evaluationId: type: number organizationId: $ref: '#/components/schemas/OrganizationId' score: type: number total: oneOf: - type: 'null' - type: number traceCoverageRate: oneOf: - type: 'null' - type: string provenanceCoverageRate: oneOf: - type: 'null' - type: string breakdown: $ref: '#/components/schemas/JsonValue' flags: $ref: '#/components/schemas/JsonValue' evidenceLevel: oneOf: - type: 'null' - type: string scoringVersion: type: string model: oneOf: - type: 'null' - type: string createdAt: type: string format: date-time additionalProperties: false required: - id - evaluationRunId - evaluationId - organizationId - score - total - traceCoverageRate - provenanceCoverageRate - breakdown - flags - evidenceLevel - scoringVersion - model - createdAt count: type: number additionalProperties: false required: - evaluationId - integrity - workflowCoverage - qualityScoreTrend - count JsonValue: oneOf: - type: 'null' - type: boolean - type: number - type: string - type: array items: $ref: '#/components/schemas/JsonValue' - type: object additionalProperties: $ref: '#/components/schemas/JsonValue' OrganizationId: type: string format: uuid ApiError: type: object properties: error: type: object properties: code: type: string enum: - UNAUTHORIZED - FORBIDDEN - NOT_FOUND - VALIDATION_ERROR - RATE_LIMITED - CONFLICT - INTERNAL_ERROR - SERVICE_UNAVAILABLE - QUOTA_EXCEEDED - NO_ORG_MEMBERSHIP - CROSS_ORG_REFERENCE - INCOMPLETE - MALFORMED message: type: string details: {} requestId: type: string format: uuid required: - code - message required: - error responses: ValidationError: description: Validation error content: application/json: schema: $ref: '#/components/schemas/ApiError' Unauthorized: description: Unauthorized content: application/json: schema: $ref: '#/components/schemas/ApiError' Forbidden: description: Forbidden content: application/json: schema: $ref: '#/components/schemas/ApiError' NotFound: description: Resource not found content: application/json: schema: $ref: '#/components/schemas/ApiError' Conflict: description: Conflict content: application/json: schema: $ref: '#/components/schemas/ApiError' RateLimited: description: Rate limit exceeded content: application/json: schema: $ref: '#/components/schemas/ApiError' InternalError: description: Internal server error content: application/json: schema: $ref: '#/components/schemas/ApiError' securitySchemes: bearerAuth: type: http scheme: bearer bearerFormat: API Key or Session Token ````