@wildo-ai/saas-specifications 1.1.9 → 1.1.11
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/esm/_catalogs/schema-org/_catalog-spec.shared.schemas.d.ts +5 -4
- package/dist/esm/_catalogs/schema-org/_catalog-spec.shared.schemas.d.ts.map +1 -1
- package/dist/esm/_catalogs/schema-org/_catalog-spec.shared.schemas.js +5 -4
- package/dist/esm/_catalogs/schema-org/_catalog-spec.shared.schemas.js.map +1 -1
- package/dist/esm/_catalogs/schema-org/index.js +1 -1
- package/dist/esm/_core/application-maturity/application-maturity.schemas.js +1 -1
- package/dist/esm/_core/application-maturity/application-maturity.schemas.js.map +1 -1
- package/dist/esm/_core/application-specification-family-identity.schemas.d.ts +11 -0
- package/dist/esm/_core/application-specification-family-identity.schemas.d.ts.map +1 -1
- package/dist/esm/_core/application-specification-family-identity.schemas.js +11 -0
- package/dist/esm/_core/application-specification-family-identity.schemas.js.map +1 -1
- package/dist/esm/_core/brand/brand-name-choice.d.ts +72 -10
- package/dist/esm/_core/brand/brand-name-choice.d.ts.map +1 -1
- package/dist/esm/_core/brand/brand-name-choice.js +144 -24
- package/dist/esm/_core/brand/brand-name-choice.js.map +1 -1
- package/dist/esm/_core/brand/brand-name-screening-generation.schemas.d.ts +41 -1
- package/dist/esm/_core/brand/brand-name-screening-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/brand/brand-name-screening-generation.schemas.js +19 -3
- package/dist/esm/_core/brand/brand-name-screening-generation.schemas.js.map +1 -1
- package/dist/esm/_core/brand/brand-name-screening-validation.d.ts +7 -1
- package/dist/esm/_core/brand/brand-name-screening-validation.d.ts.map +1 -1
- package/dist/esm/_core/brand/brand-name-screening-validation.js +24 -3
- package/dist/esm/_core/brand/brand-name-screening-validation.js.map +1 -1
- package/dist/esm/_core/brand/brand-name-screening-view.d.ts +61 -9
- package/dist/esm/_core/brand/brand-name-screening-view.d.ts.map +1 -1
- package/dist/esm/_core/brand/brand-name-screening-view.js +42 -1
- package/dist/esm/_core/brand/brand-name-screening-view.js.map +1 -1
- package/dist/esm/_core/brand/brand-name-screening.schemas.d.ts +130 -2
- package/dist/esm/_core/brand/brand-name-screening.schemas.d.ts.map +1 -1
- package/dist/esm/_core/brand/brand-name-screening.schemas.js +165 -3
- package/dist/esm/_core/brand/brand-name-screening.schemas.js.map +1 -1
- package/dist/esm/_core/brand/brand-visual-tone.schemas.d.ts +25 -7
- package/dist/esm/_core/brand/brand-visual-tone.schemas.d.ts.map +1 -1
- package/dist/esm/_core/brand/brand-visual-tone.schemas.js +26 -8
- package/dist/esm/_core/brand/brand-visual-tone.schemas.js.map +1 -1
- package/dist/esm/_core/build-plan/build-plan-task-graph-generation.schemas.d.ts +76 -0
- package/dist/esm/_core/build-plan/build-plan-task-graph-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/build-plan/build-plan-task-graph-validation.d.ts +7 -2
- package/dist/esm/_core/build-plan/build-plan-task-graph-validation.d.ts.map +1 -1
- package/dist/esm/_core/build-plan/build-plan-task-graph-validation.js +21 -3
- package/dist/esm/_core/build-plan/build-plan-task-graph-validation.js.map +1 -1
- package/dist/esm/_core/build-plan/build-plan-task-graph.schemas.d.ts +411 -13
- package/dist/esm/_core/build-plan/build-plan-task-graph.schemas.d.ts.map +1 -1
- package/dist/esm/_core/build-plan/build-plan-task-graph.schemas.js +281 -8
- package/dist/esm/_core/build-plan/build-plan-task-graph.schemas.js.map +1 -1
- package/dist/esm/_core/build-plan/build-plan.family-aspects.js +1 -1
- package/dist/esm/_core/build-plan/build-plan.family-aspects.js.map +1 -1
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot-generation.schemas.d.ts +4 -0
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot-generation.schemas.js +3 -0
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot-generation.schemas.js.map +1 -1
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot-validation.d.ts +6 -0
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot-validation.d.ts.map +1 -1
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot-validation.js +20 -0
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot-validation.js.map +1 -1
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot.schemas.d.ts +1 -0
- package/dist/esm/_core/business-model-lite/business-model-lite-snapshot.schemas.d.ts.map +1 -1
- package/dist/esm/_core/business-model-lite/pricing-plan.schemas.d.ts +1 -0
- package/dist/esm/_core/business-model-lite/pricing-plan.schemas.d.ts.map +1 -1
- package/dist/esm/_core/business-model-lite/pricing-plan.schemas.js +7 -0
- package/dist/esm/_core/business-model-lite/pricing-plan.schemas.js.map +1 -1
- package/dist/esm/_core/capability-composition/capability-composition.schemas.d.ts +4 -3
- package/dist/esm/_core/capability-composition/capability-composition.schemas.d.ts.map +1 -1
- package/dist/esm/_core/capability-composition/capability-composition.schemas.js +4 -3
- package/dist/esm/_core/capability-composition/capability-composition.schemas.js.map +1 -1
- package/dist/esm/_core/compliance-documents/ai-document-facts.schemas.d.ts +265 -0
- package/dist/esm/_core/compliance-documents/ai-document-facts.schemas.d.ts.map +1 -0
- package/dist/esm/_core/compliance-documents/ai-document-facts.schemas.js +371 -0
- package/dist/esm/_core/compliance-documents/ai-document-facts.schemas.js.map +1 -0
- package/dist/esm/_core/compliance-documents/computed-document-facts.schemas.d.ts +55 -5
- package/dist/esm/_core/compliance-documents/computed-document-facts.schemas.d.ts.map +1 -1
- package/dist/esm/_core/compliance-documents/computed-document-facts.schemas.js +66 -2
- package/dist/esm/_core/compliance-documents/computed-document-facts.schemas.js.map +1 -1
- package/dist/esm/_core/compliance-documents/launch-blockers.schemas.d.ts +42 -2
- package/dist/esm/_core/compliance-documents/launch-blockers.schemas.d.ts.map +1 -1
- package/dist/esm/_core/compliance-documents/launch-blockers.schemas.js +174 -1
- package/dist/esm/_core/compliance-documents/launch-blockers.schemas.js.map +1 -1
- package/dist/esm/_core/compliance-documents/processor-register.schemas.d.ts +4 -1
- package/dist/esm/_core/compliance-documents/processor-register.schemas.d.ts.map +1 -1
- package/dist/esm/_core/compliance-documents/processor-register.schemas.js +10 -1
- package/dist/esm/_core/compliance-documents/processor-register.schemas.js.map +1 -1
- package/dist/esm/_core/compliance-documents/recipient-disclosures.schemas.d.ts +102 -10
- package/dist/esm/_core/compliance-documents/recipient-disclosures.schemas.d.ts.map +1 -1
- package/dist/esm/_core/compliance-documents/recipient-disclosures.schemas.js +128 -25
- package/dist/esm/_core/compliance-documents/recipient-disclosures.schemas.js.map +1 -1
- package/dist/esm/_core/compliance-documents/recipient-population.schemas.d.ts +28 -0
- package/dist/esm/_core/compliance-documents/recipient-population.schemas.d.ts.map +1 -1
- package/dist/esm/_core/compliance-documents/recipient-population.schemas.js +36 -1
- package/dist/esm/_core/compliance-documents/recipient-population.schemas.js.map +1 -1
- package/dist/esm/_core/compliance-documents/required-document-set.schemas.d.ts +15 -0
- package/dist/esm/_core/compliance-documents/required-document-set.schemas.d.ts.map +1 -1
- package/dist/esm/_core/compliance-documents/required-document-set.schemas.js +15 -0
- package/dist/esm/_core/compliance-documents/required-document-set.schemas.js.map +1 -1
- package/dist/esm/_core/compliance-documents/subject-data-recipients.schemas.d.ts +21 -10
- package/dist/esm/_core/compliance-documents/subject-data-recipients.schemas.d.ts.map +1 -1
- package/dist/esm/_core/compliance-documents/subject-data-recipients.schemas.js +130 -14
- package/dist/esm/_core/compliance-documents/subject-data-recipients.schemas.js.map +1 -1
- package/dist/esm/_core/data-seed-specification.schemas.d.ts.map +1 -1
- package/dist/esm/_core/data-seed-specification.schemas.js +5 -1
- package/dist/esm/_core/data-seed-specification.schemas.js.map +1 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set-generation.schemas.d.ts +32 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set-generation.schemas.js +41 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set-generation.schemas.js.map +1 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set-validation.d.ts +33 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set-validation.d.ts.map +1 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set-validation.js +134 -9
- package/dist/esm/_core/domain-plan/domain-plan-output-set-validation.js.map +1 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set.schemas.d.ts +117 -0
- package/dist/esm/_core/domain-plan/domain-plan-output-set.schemas.d.ts.map +1 -1
- package/dist/esm/_core/domain-plan/domain-plan-output-set.schemas.js +118 -0
- package/dist/esm/_core/domain-plan/domain-plan-output-set.schemas.js.map +1 -1
- package/dist/esm/_core/domain-plan/domain-plan.family-aspects.d.ts +1 -0
- package/dist/esm/_core/domain-plan/domain-plan.family-aspects.d.ts.map +1 -1
- package/dist/esm/_core/domain-plan/domain-plan.family-aspects.js +22 -1
- package/dist/esm/_core/domain-plan/domain-plan.family-aspects.js.map +1 -1
- package/dist/esm/_core/experience-plan/experience-plan-generation.schemas.d.ts +17 -1
- package/dist/esm/_core/experience-plan/experience-plan-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/experience-plan/experience-plan-generation.schemas.js +21 -5
- package/dist/esm/_core/experience-plan/experience-plan-generation.schemas.js.map +1 -1
- package/dist/esm/_core/experience-plan/experience-plan-validation.d.ts +7 -1
- package/dist/esm/_core/experience-plan/experience-plan-validation.d.ts.map +1 -1
- package/dist/esm/_core/experience-plan/experience-plan-validation.js +39 -1
- package/dist/esm/_core/experience-plan/experience-plan-validation.js.map +1 -1
- package/dist/esm/_core/experience-plan/experience-plan.family-aspects.d.ts.map +1 -1
- package/dist/esm/_core/experience-plan/experience-plan.family-aspects.js +31 -5
- package/dist/esm/_core/experience-plan/experience-plan.family-aspects.js.map +1 -1
- package/dist/esm/_core/experience-plan/experience-plan.schemas.d.ts +81 -9
- package/dist/esm/_core/experience-plan/experience-plan.schemas.d.ts.map +1 -1
- package/dist/esm/_core/experience-plan/experience-plan.schemas.js +88 -7
- package/dist/esm/_core/experience-plan/experience-plan.schemas.js.map +1 -1
- package/dist/esm/_core/expert-personas/expert-persona-charter.schemas.d.ts +13 -6
- package/dist/esm/_core/expert-personas/expert-persona-charter.schemas.d.ts.map +1 -1
- package/dist/esm/_core/expert-personas/expert-persona-charter.schemas.js +13 -6
- package/dist/esm/_core/expert-personas/expert-persona-charter.schemas.js.map +1 -1
- package/dist/esm/_core/family-aspects.registry.d.ts.map +1 -1
- package/dist/esm/_core/family-aspects.registry.js +2 -0
- package/dist/esm/_core/family-aspects.registry.js.map +1 -1
- package/dist/esm/_core/family-aspects.schemas.d.ts +24 -0
- package/dist/esm/_core/family-aspects.schemas.d.ts.map +1 -1
- package/dist/esm/_core/family-aspects.schemas.js +20 -0
- package/dist/esm/_core/family-aspects.schemas.js.map +1 -1
- package/dist/esm/_core/feature-drafts/capability-map-output-set-generation.schemas.d.ts +6 -6
- package/dist/esm/_core/feature-drafts/feature-specification-drafts.schemas.js +2 -2
- package/dist/esm/_core/feature-drafts/feature-specification-drafts.schemas.js.map +1 -1
- package/dist/esm/_core/index.d.ts +6 -0
- package/dist/esm/_core/index.d.ts.map +1 -1
- package/dist/esm/_core/index.js +6 -0
- package/dist/esm/_core/index.js.map +1 -1
- package/dist/esm/_core/integration-plan/integration-plan-generation.schemas.d.ts +1 -0
- package/dist/esm/_core/integration-plan/integration-plan-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/integration-plan/integration-plan.family-aspects.d.ts.map +1 -1
- package/dist/esm/_core/integration-plan/integration-plan.family-aspects.js +2 -1
- package/dist/esm/_core/integration-plan/integration-plan.family-aspects.js.map +1 -1
- package/dist/esm/_core/integration-plan/integration-plan.schemas.d.ts +26 -2
- package/dist/esm/_core/integration-plan/integration-plan.schemas.d.ts.map +1 -1
- package/dist/esm/_core/integration-plan/integration-plan.schemas.js +21 -1
- package/dist/esm/_core/integration-plan/integration-plan.schemas.js.map +1 -1
- package/dist/esm/_core/market/market-references.schemas.d.ts +7 -0
- package/dist/esm/_core/market/market-references.schemas.d.ts.map +1 -1
- package/dist/esm/_core/market/market-references.schemas.js +13 -1
- package/dist/esm/_core/market/market-references.schemas.js.map +1 -1
- package/dist/esm/_core/market/market-snapshot-generation.schemas.d.ts +18 -12
- package/dist/esm/_core/market/market-snapshot-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/market/market-snapshot-generation.schemas.js +46 -17
- package/dist/esm/_core/market/market-snapshot-generation.schemas.js.map +1 -1
- package/dist/esm/_core/market/market.family-aspects.d.ts.map +1 -1
- package/dist/esm/_core/market/market.family-aspects.js +4 -1
- package/dist/esm/_core/market/market.family-aspects.js.map +1 -1
- package/dist/esm/_core/placement-plan/placement-plan-generation.schemas.d.ts +15 -3
- package/dist/esm/_core/placement-plan/placement-plan-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/placement-plan/placement-plan-generation.schemas.js +6 -3
- package/dist/esm/_core/placement-plan/placement-plan-generation.schemas.js.map +1 -1
- package/dist/esm/_core/placement-plan/placement-plan-validation.d.ts +12 -1
- package/dist/esm/_core/placement-plan/placement-plan-validation.d.ts.map +1 -1
- package/dist/esm/_core/placement-plan/placement-plan-validation.js +56 -3
- package/dist/esm/_core/placement-plan/placement-plan-validation.js.map +1 -1
- package/dist/esm/_core/placement-plan/placement-plan.family-aspects.d.ts.map +1 -1
- package/dist/esm/_core/placement-plan/placement-plan.family-aspects.js +9 -0
- package/dist/esm/_core/placement-plan/placement-plan.family-aspects.js.map +1 -1
- package/dist/esm/_core/placement-plan/placement-plan.schemas.d.ts +83 -0
- package/dist/esm/_core/placement-plan/placement-plan.schemas.d.ts.map +1 -1
- package/dist/esm/_core/placement-plan/placement-plan.schemas.js +102 -2
- package/dist/esm/_core/placement-plan/placement-plan.schemas.js.map +1 -1
- package/dist/esm/_core/requirements/functional-requirement.schemas.d.ts +3 -2
- package/dist/esm/_core/requirements/functional-requirement.schemas.d.ts.map +1 -1
- package/dist/esm/_core/requirements/functional-requirement.schemas.js +8 -2
- package/dist/esm/_core/requirements/functional-requirement.schemas.js.map +1 -1
- package/dist/esm/_core/requirements/non-functional-requirement.schemas.d.ts +1 -0
- package/dist/esm/_core/requirements/non-functional-requirement.schemas.d.ts.map +1 -1
- package/dist/esm/_core/requirements/non-functional-requirement.schemas.js +2 -0
- package/dist/esm/_core/requirements/non-functional-requirement.schemas.js.map +1 -1
- package/dist/esm/_core/requirements/requirements-pain-coverage.d.ts +41 -0
- package/dist/esm/_core/requirements/requirements-pain-coverage.d.ts.map +1 -0
- package/dist/esm/_core/requirements/requirements-pain-coverage.js +63 -0
- package/dist/esm/_core/requirements/requirements-pain-coverage.js.map +1 -0
- package/dist/esm/_core/requirements/requirements-references.schemas.d.ts +1 -1
- package/dist/esm/_core/requirements/requirements-references.schemas.js +1 -1
- package/dist/esm/_core/requirements/requirements-references.schemas.js.map +1 -1
- package/dist/esm/_core/requirements/requirements-snapshot-generation.schemas.d.ts +16 -0
- package/dist/esm/_core/requirements/requirements-snapshot-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/requirements/requirements-snapshot-generation.schemas.js +12 -1
- package/dist/esm/_core/requirements/requirements-snapshot-generation.schemas.js.map +1 -1
- package/dist/esm/_core/requirements/requirements-snapshot-validation.d.ts +13 -2
- package/dist/esm/_core/requirements/requirements-snapshot-validation.d.ts.map +1 -1
- package/dist/esm/_core/requirements/requirements-snapshot-validation.js +32 -1
- package/dist/esm/_core/requirements/requirements-snapshot-validation.js.map +1 -1
- package/dist/esm/_core/requirements/requirements-snapshot.schemas.d.ts +65 -1
- package/dist/esm/_core/requirements/requirements-snapshot.schemas.d.ts.map +1 -1
- package/dist/esm/_core/requirements/requirements-snapshot.schemas.js +53 -1
- package/dist/esm/_core/requirements/requirements-snapshot.schemas.js.map +1 -1
- package/dist/esm/_core/requirements/requirements.family-aspects.d.ts.map +1 -1
- package/dist/esm/_core/requirements/requirements.family-aspects.js +2 -0
- package/dist/esm/_core/requirements/requirements.family-aspects.js.map +1 -1
- package/dist/esm/_core/resource-specification.schemas.d.ts +56 -15
- package/dist/esm/_core/resource-specification.schemas.d.ts.map +1 -1
- package/dist/esm/_core/resource-specification.schemas.js +73 -13
- package/dist/esm/_core/resource-specification.schemas.js.map +1 -1
- package/dist/esm/_core/roadmap/roadmap-phase.schemas.d.ts +1 -16
- package/dist/esm/_core/roadmap/roadmap-phase.schemas.d.ts.map +1 -1
- package/dist/esm/_core/roadmap/roadmap-phase.schemas.js +3 -17
- package/dist/esm/_core/roadmap/roadmap-phase.schemas.js.map +1 -1
- package/dist/esm/_core/roadmap/roadmap-snapshot-generation.schemas.d.ts +2 -1
- package/dist/esm/_core/roadmap/roadmap-snapshot-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/roadmap/roadmap-snapshot-generation.schemas.js +2 -1
- package/dist/esm/_core/roadmap/roadmap-snapshot-generation.schemas.js.map +1 -1
- package/dist/esm/_core/roadmap/roadmap-snapshot.schemas.d.ts +1 -1
- package/dist/esm/_core/service-method/service-method-derivation.d.ts +155 -0
- package/dist/esm/_core/service-method/service-method-derivation.d.ts.map +1 -0
- package/dist/esm/_core/service-method/service-method-derivation.js +410 -0
- package/dist/esm/_core/service-method/service-method-derivation.js.map +1 -0
- package/dist/esm/_core/service-method/service-method-generation.schemas.d.ts +86 -0
- package/dist/esm/_core/service-method/service-method-generation.schemas.d.ts.map +1 -0
- package/dist/esm/_core/service-method/service-method-generation.schemas.js +28 -0
- package/dist/esm/_core/service-method/service-method-generation.schemas.js.map +1 -0
- package/dist/esm/_core/service-method/service-method-validation.d.ts +35 -0
- package/dist/esm/_core/service-method/service-method-validation.d.ts.map +1 -0
- package/dist/esm/_core/service-method/service-method-validation.js +75 -0
- package/dist/esm/_core/service-method/service-method-validation.js.map +1 -0
- package/dist/esm/_core/service-method/service-method.family-aspects.d.ts +3 -0
- package/dist/esm/_core/service-method/service-method.family-aspects.d.ts.map +1 -0
- package/dist/esm/_core/service-method/service-method.family-aspects.js +59 -0
- package/dist/esm/_core/service-method/service-method.family-aspects.js.map +1 -0
- package/dist/esm/_core/service-method/service-method.schemas.d.ts +385 -0
- package/dist/esm/_core/service-method/service-method.schemas.d.ts.map +1 -0
- package/dist/esm/_core/service-method/service-method.schemas.js +372 -0
- package/dist/esm/_core/service-method/service-method.schemas.js.map +1 -0
- package/dist/esm/_core/website-plan/website-plan-generation.schemas.d.ts +37 -5
- package/dist/esm/_core/website-plan/website-plan-generation.schemas.d.ts.map +1 -1
- package/dist/esm/_core/website-plan/website-plan-generation.schemas.js +34 -6
- package/dist/esm/_core/website-plan/website-plan-generation.schemas.js.map +1 -1
- package/dist/esm/_core/website-plan/website-plan-validation.d.ts +29 -3
- package/dist/esm/_core/website-plan/website-plan-validation.d.ts.map +1 -1
- package/dist/esm/_core/website-plan/website-plan-validation.js +97 -5
- package/dist/esm/_core/website-plan/website-plan-validation.js.map +1 -1
- package/dist/esm/_core/website-plan/website-plan.family-aspects.d.ts.map +1 -1
- package/dist/esm/_core/website-plan/website-plan.family-aspects.js +6 -1
- package/dist/esm/_core/website-plan/website-plan.family-aspects.js.map +1 -1
- package/dist/esm/_core/website-plan/website-plan.schemas.d.ts +521 -15
- package/dist/esm/_core/website-plan/website-plan.schemas.d.ts.map +1 -1
- package/dist/esm/_core/website-plan/website-plan.schemas.js +703 -23
- package/dist/esm/_core/website-plan/website-plan.schemas.js.map +1 -1
- package/dist/esm/application-creation/accepted-change-attribution.application-creation.schemas.d.ts +78 -0
- package/dist/esm/application-creation/accepted-change-attribution.application-creation.schemas.d.ts.map +1 -0
- package/dist/esm/application-creation/accepted-change-attribution.application-creation.schemas.js +92 -0
- package/dist/esm/application-creation/accepted-change-attribution.application-creation.schemas.js.map +1 -0
- package/dist/esm/application-creation/application-creator-inputs.application-creation.schemas.d.ts +26 -0
- package/dist/esm/application-creation/application-creator-inputs.application-creation.schemas.d.ts.map +1 -1
- package/dist/esm/application-creation/application-creator-inputs.application-creation.schemas.js +14 -0
- package/dist/esm/application-creation/application-creator-inputs.application-creation.schemas.js.map +1 -1
- package/dist/esm/application-creation/application-document-fact-specification.application-creation.schemas.d.ts +21 -5
- package/dist/esm/application-creation/application-document-fact-specification.application-creation.schemas.d.ts.map +1 -1
- package/dist/esm/application-creation/application-document-fact-specification.application-creation.schemas.js +22 -4
- package/dist/esm/application-creation/application-document-fact-specification.application-creation.schemas.js.map +1 -1
- package/dist/esm/application-creation/application-line.application-creation.schemas.d.ts +119 -10
- package/dist/esm/application-creation/application-line.application-creation.schemas.d.ts.map +1 -1
- package/dist/esm/application-creation/application-line.application-creation.schemas.js +140 -7
- package/dist/esm/application-creation/application-line.application-creation.schemas.js.map +1 -1
- package/dist/esm/application-creation/domain-plan-evaluation-identities.application-creation.definitions.d.ts +7 -1
- package/dist/esm/application-creation/domain-plan-evaluation-identities.application-creation.definitions.d.ts.map +1 -1
- package/dist/esm/application-creation/domain-plan-evaluation-identities.application-creation.definitions.js +7 -1
- package/dist/esm/application-creation/domain-plan-evaluation-identities.application-creation.definitions.js.map +1 -1
- package/dist/esm/application-creation/journey.application-creation.schemas.d.ts +4 -2
- package/dist/esm/application-creation/journey.application-creation.schemas.d.ts.map +1 -1
- package/dist/esm/application-creation/journey.application-creation.schemas.js +4 -2
- package/dist/esm/application-creation/journey.application-creation.schemas.js.map +1 -1
- package/dist/esm/application-creation/playbooks.application-creation.schemas.d.ts +103 -15
- package/dist/esm/application-creation/playbooks.application-creation.schemas.d.ts.map +1 -1
- package/dist/esm/application-creation/playbooks.application-creation.schemas.js +151 -21
- package/dist/esm/application-creation/playbooks.application-creation.schemas.js.map +1 -1
- package/dist/esm/application-creation/structured-artifact-landing.application-creation.schemas.d.ts +1 -0
- package/dist/esm/application-creation/structured-artifact-landing.application-creation.schemas.d.ts.map +1 -1
- package/dist/esm/application-creation/structured-artifact-landing.application-creation.schemas.js +1 -0
- package/dist/esm/application-creation/structured-artifact-landing.application-creation.schemas.js.map +1 -1
- package/dist/esm/application-creation.exports.d.ts +7 -0
- package/dist/esm/application-creation.exports.d.ts.map +1 -1
- package/dist/esm/application-creation.exports.js +7 -0
- package/dist/esm/application-creation.exports.js.map +1 -1
- package/dist/esm/assurance/application-risk-signals.derivations.d.ts +36 -2
- package/dist/esm/assurance/application-risk-signals.derivations.d.ts.map +1 -1
- package/dist/esm/assurance/application-risk-signals.derivations.js +52 -4
- package/dist/esm/assurance/application-risk-signals.derivations.js.map +1 -1
- package/dist/esm/assurance/assurance-assessment-twins.d.ts +38 -0
- package/dist/esm/assurance/assurance-assessment-twins.d.ts.map +1 -0
- package/dist/esm/assurance/assurance-assessment-twins.js +36 -0
- package/dist/esm/assurance/assurance-assessment-twins.js.map +1 -0
- package/dist/esm/assurance/assurance-authority-basis.definitions.d.ts +608 -9
- package/dist/esm/assurance/assurance-authority-basis.definitions.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-authority-basis.definitions.js +2962 -1078
- package/dist/esm/assurance/assurance-authority-basis.definitions.js.map +1 -1
- package/dist/esm/assurance/assurance-authority-basis.schemas.d.ts +15 -0
- package/dist/esm/assurance/assurance-authority-basis.schemas.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-authority-basis.schemas.js +1 -0
- package/dist/esm/assurance/assurance-authority-basis.schemas.js.map +1 -1
- package/dist/esm/assurance/assurance-document-system.definitions.d.ts +2 -0
- package/dist/esm/assurance/assurance-document-system.definitions.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-document-system.definitions.js +138 -19
- package/dist/esm/assurance/assurance-document-system.definitions.js.map +1 -1
- package/dist/esm/assurance/assurance-document-system.schemas.d.ts +65 -0
- package/dist/esm/assurance/assurance-document-system.schemas.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-document-system.schemas.js +13 -0
- package/dist/esm/assurance/assurance-document-system.schemas.js.map +1 -1
- package/dist/esm/assurance/assurance-evidence-judgement.schemas.d.ts +81 -6
- package/dist/esm/assurance/assurance-evidence-judgement.schemas.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-evidence-judgement.schemas.js +68 -7
- package/dist/esm/assurance/assurance-evidence-judgement.schemas.js.map +1 -1
- package/dist/esm/assurance/assurance-evidence-requirement-bodies.definitions.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-evidence-requirement-bodies.definitions.js +1631 -70
- package/dist/esm/assurance/assurance-evidence-requirement-bodies.definitions.js.map +1 -1
- package/dist/esm/assurance/assurance-host-implementation-artifacts.d.ts +77 -0
- package/dist/esm/assurance/assurance-host-implementation-artifacts.d.ts.map +1 -0
- package/dist/esm/assurance/assurance-host-implementation-artifacts.js +51 -0
- package/dist/esm/assurance/assurance-host-implementation-artifacts.js.map +1 -0
- package/dist/esm/assurance/assurance-objective-catalog.definitions.d.ts +3 -2
- package/dist/esm/assurance/assurance-objective-catalog.definitions.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-objective-catalog.definitions.js +327 -3
- package/dist/esm/assurance/assurance-objective-catalog.definitions.js.map +1 -1
- package/dist/esm/assurance/assurance-open-compliance-answer-cell.d.ts +12 -0
- package/dist/esm/assurance/assurance-open-compliance-answer-cell.d.ts.map +1 -0
- package/dist/esm/assurance/assurance-open-compliance-answer-cell.js +19 -0
- package/dist/esm/assurance/assurance-open-compliance-answer-cell.js.map +1 -0
- package/dist/esm/assurance/assurance-open-compliance-coverage-pack.d.ts +403 -0
- package/dist/esm/assurance/assurance-open-compliance-coverage-pack.d.ts.map +1 -0
- package/dist/esm/assurance/assurance-open-compliance-coverage-pack.js +825 -0
- package/dist/esm/assurance/assurance-open-compliance-coverage-pack.js.map +1 -0
- package/dist/esm/assurance/assurance-program-generation.schemas.d.ts +83 -8
- package/dist/esm/assurance/assurance-program-generation.schemas.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-program-generation.schemas.js +167 -20
- package/dist/esm/assurance/assurance-program-generation.schemas.js.map +1 -1
- package/dist/esm/assurance/assurance-program.schemas.d.ts +234 -35
- package/dist/esm/assurance/assurance-program.schemas.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-program.schemas.js +255 -14
- package/dist/esm/assurance/assurance-program.schemas.js.map +1 -1
- package/dist/esm/assurance/assurance-source-registry.definitions.d.ts +2 -1
- package/dist/esm/assurance/assurance-source-registry.definitions.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-source-registry.definitions.js +259 -18
- package/dist/esm/assurance/assurance-source-registry.definitions.js.map +1 -1
- package/dist/esm/assurance/assurance-source-requirement-inventory.definitions.d.ts +17 -0
- package/dist/esm/assurance/assurance-source-requirement-inventory.definitions.d.ts.map +1 -0
- package/dist/esm/assurance/assurance-source-requirement-inventory.definitions.js +216 -0
- package/dist/esm/assurance/assurance-source-requirement-inventory.definitions.js.map +1 -0
- package/dist/esm/assurance/assurance-sources.schemas.d.ts +110 -0
- package/dist/esm/assurance/assurance-sources.schemas.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-sources.schemas.js +111 -1
- package/dist/esm/assurance/assurance-sources.schemas.js.map +1 -1
- package/dist/esm/assurance/assurance-verification-requirement-bodies.definitions.d.ts.map +1 -1
- package/dist/esm/assurance/assurance-verification-requirement-bodies.definitions.js +356 -5
- package/dist/esm/assurance/assurance-verification-requirement-bodies.definitions.js.map +1 -1
- package/dist/esm/assurance/governed-document-audience.derivations.d.ts +83 -0
- package/dist/esm/assurance/governed-document-audience.derivations.d.ts.map +1 -0
- package/dist/esm/assurance/governed-document-audience.derivations.js +97 -0
- package/dist/esm/assurance/governed-document-audience.derivations.js.map +1 -0
- package/dist/esm/assurance/governed-document-release.d.ts +15 -1
- package/dist/esm/assurance/governed-document-release.d.ts.map +1 -1
- package/dist/esm/assurance/governed-document-release.js +3 -1
- package/dist/esm/assurance/governed-document-release.js.map +1 -1
- package/dist/esm/assurance.exports.d.ts +4 -0
- package/dist/esm/assurance.exports.d.ts.map +1 -1
- package/dist/esm/assurance.exports.js +4 -0
- package/dist/esm/assurance.exports.js.map +1 -1
- package/dist/esm/companion-exports.d.ts +9 -1
- package/dist/esm/companion-exports.d.ts.map +1 -1
- package/dist/esm/companion-exports.js.map +1 -1
- package/dist/esm/engine/applications/resources/application-api-keys.resource.specification.d.ts +1 -1
- package/dist/esm/engine/applications/resources/application-api-keys.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/applications/resources/application-api-keys.resource.specification.js +33 -5
- package/dist/esm/engine/applications/resources/application-api-keys.resource.specification.js.map +1 -1
- package/dist/esm/engine/applications/resources/application-metadata.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/applications/resources/application-metadata.resource.specification.js +9 -12
- package/dist/esm/engine/applications/resources/application-metadata.resource.specification.js.map +1 -1
- package/dist/esm/engine/applications/resources/application-oauth-clients.resource.specification.d.ts +1 -1
- package/dist/esm/engine/applications/resources/application-oauth-clients.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/applications/resources/application-oauth-clients.resource.specification.js +60 -6
- package/dist/esm/engine/applications/resources/application-oauth-clients.resource.specification.js.map +1 -1
- package/dist/esm/engine/applications/resources/application-versions.resource.specification.js +4 -4
- package/dist/esm/engine/applications/resources/application-versions.resource.specification.js.map +1 -1
- package/dist/esm/engine/applications/resources/applications.resource.specification.d.ts +1 -0
- package/dist/esm/engine/applications/resources/applications.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/applications/resources/applications.resource.specification.js +2 -2
- package/dist/esm/engine/applications/resources/applications.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/application-access-grants.resource.specification.d.ts +12 -4
- package/dist/esm/engine/auth/resources/application-access-grants.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/application-access-grants.resource.specification.js +46 -10
- package/dist/esm/engine/auth/resources/application-access-grants.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/oauth-consent-grants.resource.specification.d.ts +3 -2
- package/dist/esm/engine/auth/resources/oauth-consent-grants.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/oauth-consent-grants.resource.specification.js +15 -4
- package/dist/esm/engine/auth/resources/oauth-consent-grants.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/organization-api-keys.resource.specification.d.ts +1 -1
- package/dist/esm/engine/auth/resources/organization-api-keys.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/organization-api-keys.resource.specification.js +40 -11
- package/dist/esm/engine/auth/resources/organization-api-keys.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/organization-oauth-clients.resource.specification.d.ts +1 -1
- package/dist/esm/engine/auth/resources/organization-oauth-clients.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/organization-oauth-clients.resource.specification.js +44 -3
- package/dist/esm/engine/auth/resources/organization-oauth-clients.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/organization-provider-credentials.resource.specification.d.ts +7 -3
- package/dist/esm/engine/auth/resources/organization-provider-credentials.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/organization-provider-credentials.resource.specification.js +40 -8
- package/dist/esm/engine/auth/resources/organization-provider-credentials.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/platform-access-grants.resource.specification.d.ts +14 -8
- package/dist/esm/engine/auth/resources/platform-access-grants.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/platform-access-grants.resource.specification.js +72 -19
- package/dist/esm/engine/auth/resources/platform-access-grants.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/scim-provisioning-config.resource.specification.d.ts +2 -2
- package/dist/esm/engine/auth/resources/scim-provisioning-config.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/scim-provisioning-config.resource.specification.js +6 -6
- package/dist/esm/engine/auth/resources/scim-provisioning-config.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/scim-tokens.resource.specification.d.ts +1 -0
- package/dist/esm/engine/auth/resources/scim-tokens.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/scim-tokens.resource.specification.js +14 -13
- package/dist/esm/engine/auth/resources/scim-tokens.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/siem-delivery-dlq.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/siem-delivery-dlq.resource.specification.js +2 -2
- package/dist/esm/engine/auth/resources/siem-delivery-dlq.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/siem-delivery-outbox.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/siem-delivery-outbox.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/sso-config.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/sso-config.resource.specification.js +68 -12
- package/dist/esm/engine/auth/resources/sso-config.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/sso-connections.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/sso-connections.resource.specification.js +75 -19
- package/dist/esm/engine/auth/resources/sso-connections.resource.specification.js.map +1 -1
- package/dist/esm/engine/auth/resources/user-provider-credentials.resource.specification.d.ts +3 -1
- package/dist/esm/engine/auth/resources/user-provider-credentials.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/auth/resources/user-provider-credentials.resource.specification.js +11 -7
- package/dist/esm/engine/auth/resources/user-provider-credentials.resource.specification.js.map +1 -1
- package/dist/esm/engine/billing/resources/billing-accounts.resource.specification.d.ts +2 -0
- package/dist/esm/engine/billing/resources/billing-accounts.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/billing/resources/billing-accounts.resource.specification.js +2 -1
- package/dist/esm/engine/billing/resources/billing-accounts.resource.specification.js.map +1 -1
- package/dist/esm/engine/billing/resources/invoice-refs.resource.specification.d.ts +4 -2
- package/dist/esm/engine/billing/resources/invoice-refs.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/billing/resources/invoice-refs.resource.specification.js +8 -7
- package/dist/esm/engine/billing/resources/invoice-refs.resource.specification.js.map +1 -1
- package/dist/esm/engine/billing/resources/prices.resource.specification.js +1 -1
- package/dist/esm/engine/billing/resources/prices.resource.specification.js.map +1 -1
- package/dist/esm/engine/billing/resources/products.resource.specification.js +1 -1
- package/dist/esm/engine/billing/resources/products.resource.specification.js.map +1 -1
- package/dist/esm/engine/billing/resources/promotions.resource.specification.js +1 -1
- package/dist/esm/engine/billing/resources/promotions.resource.specification.js.map +1 -1
- package/dist/esm/engine/billing/resources/referrals.resource.specification.js +6 -6
- package/dist/esm/engine/billing/resources/referrals.resource.specification.js.map +1 -1
- package/dist/esm/engine/billing/resources/subscriptions.resource.specification.d.ts +2 -2
- package/dist/esm/engine/billing/resources/subscriptions.resource.specification.js +1 -1
- package/dist/esm/engine/billing/resources/subscriptions.resource.specification.js.map +1 -1
- package/dist/esm/engine/billing/resources/usage-records.resource.specification.d.ts +4 -0
- package/dist/esm/engine/billing/resources/usage-records.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/billing/resources/usage-records.resource.specification.js +4 -1
- package/dist/esm/engine/billing/resources/usage-records.resource.specification.js.map +1 -1
- package/dist/esm/engine/compliance/resources/application-audit-logs.resource.specification.d.ts +9 -5
- package/dist/esm/engine/compliance/resources/application-audit-logs.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/compliance/resources/application-audit-logs.resource.specification.js +16 -8
- package/dist/esm/engine/compliance/resources/application-audit-logs.resource.specification.js.map +1 -1
- package/dist/esm/engine/compliance/resources/approval-seals.resource.specification.d.ts +73 -0
- package/dist/esm/engine/compliance/resources/approval-seals.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/compliance/resources/approval-seals.resource.specification.js +192 -0
- package/dist/esm/engine/compliance/resources/approval-seals.resource.specification.js.map +1 -0
- package/dist/esm/engine/compliance/resources/audit-logs.resource.specification.d.ts +5 -3
- package/dist/esm/engine/compliance/resources/audit-logs.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/compliance/resources/audit-logs.resource.specification.js +19 -13
- package/dist/esm/engine/compliance/resources/audit-logs.resource.specification.js.map +1 -1
- package/dist/esm/engine/compliance/resources/legal-holds.resource.specification.d.ts +11 -6
- package/dist/esm/engine/compliance/resources/legal-holds.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/compliance/resources/legal-holds.resource.specification.js +14 -3
- package/dist/esm/engine/compliance/resources/legal-holds.resource.specification.js.map +1 -1
- package/dist/esm/engine/compliance/resources/published-verification-keys.resource.specification.d.ts +37 -0
- package/dist/esm/engine/compliance/resources/published-verification-keys.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/compliance/resources/published-verification-keys.resource.specification.js +161 -0
- package/dist/esm/engine/compliance/resources/published-verification-keys.resource.specification.js.map +1 -0
- package/dist/esm/engine/compliance/resources/retention-enforcement-records.resource.specification.js +1 -1
- package/dist/esm/engine/compliance/resources/retention-enforcement-records.resource.specification.js.map +1 -1
- package/dist/esm/engine/engine-resource-specifications.registry.d.ts.map +1 -1
- package/dist/esm/engine/engine-resource-specifications.registry.js +55 -11
- package/dist/esm/engine/engine-resource-specifications.registry.js.map +1 -1
- package/dist/esm/engine/errors/error-messages.specification.d.ts.map +1 -1
- package/dist/esm/engine/errors/error-messages.specification.js +300 -5
- package/dist/esm/engine/errors/error-messages.specification.js.map +1 -1
- package/dist/esm/engine/features/core-limits.feature.specifications.d.ts +10 -4
- package/dist/esm/engine/features/core-limits.feature.specifications.d.ts.map +1 -1
- package/dist/esm/engine/features/core-limits.feature.specifications.js +100 -5
- package/dist/esm/engine/features/core-limits.feature.specifications.js.map +1 -1
- package/dist/esm/engine/features/custom-branding.feature.specification.d.ts.map +1 -1
- package/dist/esm/engine/features/custom-branding.feature.specification.js +7 -3
- package/dist/esm/engine/features/custom-branding.feature.specification.js.map +1 -1
- package/dist/esm/engine/features/index.d.ts +4 -2
- package/dist/esm/engine/features/index.d.ts.map +1 -1
- package/dist/esm/engine/features/index.js +6 -4
- package/dist/esm/engine/features/index.js.map +1 -1
- package/dist/esm/engine/features/resources/application-features.resource.specification.js +2 -2
- package/dist/esm/engine/features/resources/application-features.resource.specification.js.map +1 -1
- package/dist/esm/engine/features/resources/organization-features.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/features/resources/organization-features.resource.specification.js +34 -15
- package/dist/esm/engine/features/resources/organization-features.resource.specification.js.map +1 -1
- package/dist/esm/engine/files/resources/files.resource.specification.d.ts +4 -1
- package/dist/esm/engine/files/resources/files.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/files/resources/files.resource.specification.js +24 -3
- package/dist/esm/engine/files/resources/files.resource.specification.js.map +1 -1
- package/dist/esm/engine/flows-actors/resources/a2a-push-delivery-log.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/flows-actors/resources/a2a-push-delivery-log.resource.specification.js +31 -3
- package/dist/esm/engine/flows-actors/resources/a2a-push-delivery-log.resource.specification.js.map +1 -1
- package/dist/esm/engine/flows-actors/resources/a2a-token-budget-state.resource.specification.js +6 -6
- package/dist/esm/engine/flows-actors/resources/a2a-token-budget-state.resource.specification.js.map +1 -1
- package/dist/esm/engine/flows-actors/resources/actor-system-executions.resource.specification.d.ts +30 -0
- package/dist/esm/engine/flows-actors/resources/actor-system-executions.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/actor-system-executions.resource.specification.js +181 -0
- package/dist/esm/engine/flows-actors/resources/actor-system-executions.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/actor-tasks.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/actor-tasks.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/actor-tasks.resource.specification.js +110 -0
- package/dist/esm/engine/flows-actors/resources/actor-tasks.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/agent-behavior-reports.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/agent-behavior-reports.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/agent-behavior-reports.resource.specification.js +286 -0
- package/dist/esm/engine/flows-actors/resources/agent-behavior-reports.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/agent-operation-records.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/agent-operation-records.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/agent-operation-records.resource.specification.js +147 -0
- package/dist/esm/engine/flows-actors/resources/agent-operation-records.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/agent-work-decisions.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/flows-actors/resources/agent-work-decisions.resource.specification.js +71 -26
- package/dist/esm/engine/flows-actors/resources/agent-work-decisions.resource.specification.js.map +1 -1
- package/dist/esm/engine/flows-actors/resources/application-agent-policies.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/application-agent-policies.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/application-agent-policies.resource.specification.js +92 -0
- package/dist/esm/engine/flows-actors/resources/application-agent-policies.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/chat-conversations.resource.specification.d.ts +46 -0
- package/dist/esm/engine/flows-actors/resources/chat-conversations.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/chat-conversations.resource.specification.js +196 -0
- package/dist/esm/engine/flows-actors/resources/chat-conversations.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/flow-actor-executions.resource.specification.d.ts +73 -31
- package/dist/esm/engine/flows-actors/resources/flow-actor-executions.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/flows-actors/resources/flow-actor-executions.resource.specification.js +118 -18
- package/dist/esm/engine/flows-actors/resources/flow-actor-executions.resource.specification.js.map +1 -1
- package/dist/esm/engine/flows-actors/resources/flow-actor-tasks.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/flow-actor-tasks.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/flows-actors/resources/flow-actor-tasks.resource.specification.js +9 -4
- package/dist/esm/engine/flows-actors/resources/flow-actor-tasks.resource.specification.js.map +1 -1
- package/dist/esm/engine/flows-actors/resources/human-decision-notice-stays.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/human-decision-notice-stays.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/human-decision-notice-stays.resource.specification.js +184 -0
- package/dist/esm/engine/flows-actors/resources/human-decision-notice-stays.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/model-call-attributions.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/flows-actors/resources/model-call-attributions.resource.specification.js +17 -5
- package/dist/esm/engine/flows-actors/resources/model-call-attributions.resource.specification.js.map +1 -1
- package/dist/esm/engine/flows-actors/resources/model-cost-period-aggregates.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/flows-actors/resources/model-cost-period-aggregates.resource.specification.js +47 -7
- package/dist/esm/engine/flows-actors/resources/model-cost-period-aggregates.resource.specification.js.map +1 -1
- package/dist/esm/engine/flows-actors/resources/model-usage-ceiling-state.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/model-usage-ceiling-state.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/model-usage-ceiling-state.resource.specification.js +118 -0
- package/dist/esm/engine/flows-actors/resources/model-usage-ceiling-state.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/operation-approval-requests.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/operation-approval-requests.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/operation-approval-requests.resource.specification.js +230 -0
- package/dist/esm/engine/flows-actors/resources/operation-approval-requests.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/organization-agent-policies.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/organization-agent-policies.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/organization-agent-policies.resource.specification.js +151 -0
- package/dist/esm/engine/flows-actors/resources/organization-agent-policies.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/organization-chat-settings.resource.specification.d.ts +23 -0
- package/dist/esm/engine/flows-actors/resources/organization-chat-settings.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/organization-chat-settings.resource.specification.js +124 -0
- package/dist/esm/engine/flows-actors/resources/organization-chat-settings.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/person-token-ceiling-state.resource.specification.d.ts +3 -0
- package/dist/esm/engine/flows-actors/resources/person-token-ceiling-state.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/flows-actors/resources/person-token-ceiling-state.resource.specification.js +114 -0
- package/dist/esm/engine/flows-actors/resources/person-token-ceiling-state.resource.specification.js.map +1 -0
- package/dist/esm/engine/flows-actors/resources/provider-call-records.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/flows-actors/resources/provider-call-records.resource.specification.js +45 -3
- package/dist/esm/engine/flows-actors/resources/provider-call-records.resource.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/actions/action-bar.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/actions/action-bar.frontend-component.specification.js +6 -0
- package/dist/esm/engine/frontend-components/actions/action-bar.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/ai-elements/chat/ai-conversation-thread.frontend-component.specification.d.ts +4 -0
- package/dist/esm/engine/frontend-components/ai-elements/chat/ai-conversation-thread.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/ai-elements/chat/ai-conversation-thread.frontend-component.specification.js +78 -0
- package/dist/esm/engine/frontend-components/ai-elements/chat/ai-conversation-thread.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/ai-elements/chat/ai-prompt-input.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/ai-elements/chat/ai-prompt-input.frontend-component.specification.js +4 -4
- package/dist/esm/engine/frontend-components/ai-elements/chat/ai-prompt-input.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/ai-elements/decision/ai-approval-queue.frontend-component.specification.js +3 -3
- package/dist/esm/engine/frontend-components/ai-elements/decision/ai-approval-queue.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/ai-elements/graph/ai-canvas.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/ai-elements/graph/ai-canvas.frontend-component.specification.js +4 -4
- package/dist/esm/engine/frontend-components/ai-elements/graph/ai-canvas.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/ai-elements/graph/ai-node.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/ai-elements/graph/ai-node.frontend-component.specification.js +3 -3
- package/dist/esm/engine/frontend-components/ai-elements/graph/ai-node.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/app/app-layout-agent-work-evaluation-export-settings.frontend-component.specification.d.ts +5 -0
- package/dist/esm/engine/frontend-components/app/app-layout-agent-work-evaluation-export-settings.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/app/app-layout-agent-work-evaluation-export-settings.frontend-component.specification.js +79 -0
- package/dist/esm/engine/frontend-components/app/app-layout-agent-work-evaluation-export-settings.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/app/app-layout-agent-work-override-rate-settings.frontend-component.specification.js +2 -2
- package/dist/esm/engine/frontend-components/app/app-layout-agent-work-override-rate-settings.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/app/app-layout-autonomous-agents-settings.frontend-component.specification.d.ts +5 -0
- package/dist/esm/engine/frontend-components/app/app-layout-autonomous-agents-settings.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/app/app-layout-autonomous-agents-settings.frontend-component.specification.js +79 -0
- package/dist/esm/engine/frontend-components/app/app-layout-autonomous-agents-settings.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/app/app-layout-connected-accounts-settings.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/app/app-layout-connected-accounts-settings.frontend-component.specification.js +3 -2
- package/dist/esm/engine/frontend-components/app/app-layout-connected-accounts-settings.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/app/app-layout-invitations-settings.frontend-component.specification.js +3 -3
- package/dist/esm/engine/frontend-components/app/app-layout-invitations-settings.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/app/app-layout-model-cost-breakdown-settings.frontend-component.specification.d.ts +5 -0
- package/dist/esm/engine/frontend-components/app/app-layout-model-cost-breakdown-settings.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/app/app-layout-model-cost-breakdown-settings.frontend-component.specification.js +79 -0
- package/dist/esm/engine/frontend-components/app/app-layout-model-cost-breakdown-settings.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/app/app-layout-representative-activity-settings.frontend-component.specification.d.ts +5 -0
- package/dist/esm/engine/frontend-components/app/app-layout-representative-activity-settings.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/app/app-layout-representative-activity-settings.frontend-component.specification.js +79 -0
- package/dist/esm/engine/frontend-components/app/app-layout-representative-activity-settings.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/app/app-layout-settings-hub.frontend-component.specification.js +3 -3
- package/dist/esm/engine/frontend-components/app/app-layout-settings-hub.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/app/public-layout-legal-notice.frontend-component.specification.d.ts +4 -0
- package/dist/esm/engine/frontend-components/app/public-layout-legal-notice.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/app/public-layout-legal-notice.frontend-component.specification.js +22 -0
- package/dist/esm/engine/frontend-components/app/public-layout-legal-notice.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/app/public-layout-main-container.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/app/public-layout-main-container.frontend-component.specification.js +2 -1
- package/dist/esm/engine/frontend-components/app/public-layout-main-container.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/auth/admin-config/app-page-security-settings.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/auth/admin-config/app-page-security-settings.frontend-component.specification.js +3 -2
- package/dist/esm/engine/frontend-components/auth/admin-config/app-page-security-settings.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/auth/public-layout-chat-identity-link.frontend-component.specification.d.ts +4 -0
- package/dist/esm/engine/frontend-components/auth/public-layout-chat-identity-link.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/auth/public-layout-chat-identity-link.frontend-component.specification.js +22 -0
- package/dist/esm/engine/frontend-components/auth/public-layout-chat-identity-link.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/auth/public-layout-user-auth.frontend-component.specification.js +2 -2
- package/dist/esm/engine/frontend-components/auth/public-layout-user-auth.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/auth/public-layout-user-invitation.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/auth/public-layout-user-invitation.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/billing/billing-pricing-table.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/billing/billing-pricing-table.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/data-display/badge.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/data-display/badge.frontend-component.specification.js +7 -0
- package/dist/esm/engine/frontend-components/data-display/badge.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/feedback/features/feature-limit-indicator.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/feedback/features/feature-limit-indicator.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-approval.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-approval.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-decision.frontend-component.specification.js +6 -6
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-decision.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-execution-usage.frontend-component.specification.d.ts +4 -0
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-execution-usage.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-execution-usage.frontend-component.specification.js +24 -0
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-execution-usage.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-flow-node.frontend-component.specification.js +6 -6
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-flow-node.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-graph-viewer.frontend-component.specification.d.ts +4 -0
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-graph-viewer.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-graph-viewer.frontend-component.specification.js +128 -0
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-graph-viewer.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-system-catalog.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/flows-actors/low-level-ai-system-catalog.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/form/controls/boolean/toggle.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/form/controls/boolean/toggle.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/date-range-picker.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/date-range-picker.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/datepicker.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/datepicker.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/datetime-picker.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/datetime-picker.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/datetime-range-picker.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/datetime-range-picker.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/time-range-picker.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/form/controls/date-time/time-range-picker.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/form/controls/enum/toggle-group.frontend-component.specification.js +4 -4
- package/dist/esm/engine/frontend-components/form/controls/enum/toggle-group.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/form/controls/string/chips-input.frontend-component.specification.d.ts +5 -0
- package/dist/esm/engine/frontend-components/form/controls/string/chips-input.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/form/controls/string/chips-input.frontend-component.specification.js +139 -0
- package/dist/esm/engine/frontend-components/form/controls/string/chips-input.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/form/fields/form-field-string.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/form/fields/form-field-string.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/high-level-views/high-level-views-flows-actors.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/high-level-views/high-level-views-flows-actors.frontend-component.specification.js +5 -1
- package/dist/esm/engine/frontend-components/high-level-views/high-level-views-flows-actors.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/index.d.ts +7 -0
- package/dist/esm/engine/frontend-components/index.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/index.js +29 -0
- package/dist/esm/engine/frontend-components/index.js.map +1 -1
- package/dist/esm/engine/frontend-components/navigation/app-level/sidebar.frontend-component.specification.js +1 -1
- package/dist/esm/engine/frontend-components/navigation/app-level/sidebar.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/navigation/app-level/user-dropdown.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/navigation/app-level/user-dropdown.frontend-component.specification.js +3 -1
- package/dist/esm/engine/frontend-components/navigation/app-level/user-dropdown.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/navigation/app-level/utility-panel.frontend-component.specification.js +4 -4
- package/dist/esm/engine/frontend-components/navigation/app-level/utility-panel.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/resources/resource-layout-header.frontend-component.specification.js +2 -2
- package/dist/esm/engine/frontend-components/resources/resource-layout-header.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/resources/resource-layout-list.frontend-component.specification.js +2 -2
- package/dist/esm/engine/frontend-components/resources/resource-layout-list.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/frontend-components/resources/resource-layout-record-sharing.frontend-component.specification.d.ts +4 -0
- package/dist/esm/engine/frontend-components/resources/resource-layout-record-sharing.frontend-component.specification.d.ts.map +1 -0
- package/dist/esm/engine/frontend-components/resources/resource-layout-record-sharing.frontend-component.specification.js +27 -0
- package/dist/esm/engine/frontend-components/resources/resource-layout-record-sharing.frontend-component.specification.js.map +1 -0
- package/dist/esm/engine/frontend-components/typography/text.frontend-component.specification.d.ts.map +1 -1
- package/dist/esm/engine/frontend-components/typography/text.frontend-component.specification.js +3 -1
- package/dist/esm/engine/frontend-components/typography/text.frontend-component.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/application-data-keys.resource.specification.d.ts +24 -0
- package/dist/esm/engine/infrastructure/resources/application-data-keys.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/infrastructure/resources/application-data-keys.resource.specification.js +136 -0
- package/dist/esm/engine/infrastructure/resources/application-data-keys.resource.specification.js.map +1 -0
- package/dist/esm/engine/infrastructure/resources/data-seed-runs.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/infrastructure/resources/data-seed-runs.resource.specification.js +6 -1
- package/dist/esm/engine/infrastructure/resources/data-seed-runs.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/document-sequence-counters.resource.specification.d.ts +9 -0
- package/dist/esm/engine/infrastructure/resources/document-sequence-counters.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/infrastructure/resources/document-sequence-counters.resource.specification.js +110 -0
- package/dist/esm/engine/infrastructure/resources/document-sequence-counters.resource.specification.js.map +1 -0
- package/dist/esm/engine/infrastructure/resources/external-data-erasure-suppressions.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-erasure-suppressions.resource.specification.js +10 -4
- package/dist/esm/engine/infrastructure/resources/external-data-erasure-suppressions.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-outbound-changes.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-outbound-changes.resource.specification.js +20 -2
- package/dist/esm/engine/infrastructure/resources/external-data-outbound-changes.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-membership-keys.resource.specification.js +6 -6
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-membership-keys.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-membership.resource.specification.d.ts +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-membership.resource.specification.js +5 -5
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-membership.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-requested-keys.resource.specification.d.ts +10 -0
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-requested-keys.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-requested-keys.resource.specification.js +118 -0
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-requested-keys.resource.specification.js.map +1 -0
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-run-records.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-run-records.resource.specification.js +54 -8
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-run-records.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-states.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-states.resource.specification.js +9 -0
- package/dist/esm/engine/infrastructure/resources/external-data-pipeline-states.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-remote-versions.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/infrastructure/resources/external-data-remote-versions.resource.specification.js +33 -5
- package/dist/esm/engine/infrastructure/resources/external-data-remote-versions.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/job-executions.resource.specification.d.ts +21 -4
- package/dist/esm/engine/infrastructure/resources/job-executions.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/infrastructure/resources/job-payload-storage.resource.specification.d.ts +3 -3
- package/dist/esm/engine/infrastructure/resources/job-payload-storage.resource.specification.js +1 -1
- package/dist/esm/engine/infrastructure/resources/job-payload-storage.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/post-commit-effect-outbox.resource.specification.d.ts +10 -0
- package/dist/esm/engine/infrastructure/resources/post-commit-effect-outbox.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/infrastructure/resources/post-commit-effect-outbox.resource.specification.js +159 -0
- package/dist/esm/engine/infrastructure/resources/post-commit-effect-outbox.resource.specification.js.map +1 -0
- package/dist/esm/engine/infrastructure/resources/system-of-record-partition-assignments.resource.specification.d.ts +5 -5
- package/dist/esm/engine/infrastructure/resources/system-of-record-partition-assignments.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/infrastructure/resources/system-of-record-partition-assignments.resource.specification.js +38 -10
- package/dist/esm/engine/infrastructure/resources/system-of-record-partition-assignments.resource.specification.js.map +1 -1
- package/dist/esm/engine/infrastructure/resources/transactional-commit-receipts.resource.specification.d.ts +9 -0
- package/dist/esm/engine/infrastructure/resources/transactional-commit-receipts.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/infrastructure/resources/transactional-commit-receipts.resource.specification.js +100 -0
- package/dist/esm/engine/infrastructure/resources/transactional-commit-receipts.resource.specification.js.map +1 -0
- package/dist/esm/engine/notifications/resources/chat-identity-links.resource.specification.d.ts +37 -0
- package/dist/esm/engine/notifications/resources/chat-identity-links.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/notifications/resources/chat-identity-links.resource.specification.js +181 -0
- package/dist/esm/engine/notifications/resources/chat-identity-links.resource.specification.js.map +1 -0
- package/dist/esm/engine/notifications/resources/conversation-threads.resource.specification.d.ts +3 -3
- package/dist/esm/engine/notifications/resources/conversation-threads.resource.specification.js +7 -7
- package/dist/esm/engine/notifications/resources/conversation-threads.resource.specification.js.map +1 -1
- package/dist/esm/engine/notifications/resources/inbound-deliveries.resource.specification.d.ts +31 -1
- package/dist/esm/engine/notifications/resources/inbound-deliveries.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/notifications/resources/inbound-deliveries.resource.specification.js +33 -7
- package/dist/esm/engine/notifications/resources/inbound-deliveries.resource.specification.js.map +1 -1
- package/dist/esm/engine/notifications/resources/inbound-email.resource.specification.d.ts +16 -14
- package/dist/esm/engine/notifications/resources/inbound-email.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/notifications/resources/inbound-email.resource.specification.js +56 -53
- package/dist/esm/engine/notifications/resources/inbound-email.resource.specification.js.map +1 -1
- package/dist/esm/engine/notifications/resources/m2m-webhook-delivery-log.resource.specification.d.ts +6 -2
- package/dist/esm/engine/notifications/resources/m2m-webhook-delivery-log.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/notifications/resources/m2m-webhook-delivery-log.resource.specification.js +22 -4
- package/dist/esm/engine/notifications/resources/m2m-webhook-delivery-log.resource.specification.js.map +1 -1
- package/dist/esm/engine/notifications/resources/notification-badges.resource.specification.d.ts +2 -2
- package/dist/esm/engine/notifications/resources/webhook-config.resource.specification.d.ts +2 -0
- package/dist/esm/engine/notifications/resources/webhook-config.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/notifications/resources/webhook-config.resource.specification.js +3 -3
- package/dist/esm/engine/notifications/resources/webhook-config.resource.specification.js.map +1 -1
- package/dist/esm/engine/notifications/system-emails/human-decision.system-emails.specification.d.ts +1 -0
- package/dist/esm/engine/notifications/system-emails/human-decision.system-emails.specification.d.ts.map +1 -1
- package/dist/esm/engine/notifications/system-emails/human-decision.system-emails.specification.js +43 -2
- package/dist/esm/engine/notifications/system-emails/human-decision.system-emails.specification.js.map +1 -1
- package/dist/esm/engine/notifications/system-emails/index.d.ts +0 -1
- package/dist/esm/engine/notifications/system-emails/index.d.ts.map +1 -1
- package/dist/esm/engine/notifications/system-emails/index.js +1 -3
- package/dist/esm/engine/notifications/system-emails/index.js.map +1 -1
- package/dist/esm/engine/notifications/system-emails/organization.system-emails.specification.d.ts +1 -1
- package/dist/esm/engine/notifications/system-emails/organization.system-emails.specification.d.ts.map +1 -1
- package/dist/esm/engine/notifications/system-emails/organization.system-emails.specification.js +25 -15
- package/dist/esm/engine/notifications/system-emails/organization.system-emails.specification.js.map +1 -1
- package/dist/esm/engine/notifications/system-emails/security.system-emails.specification.js +13 -13
- package/dist/esm/engine/notifications/system-emails/security.system-emails.specification.js.map +1 -1
- package/dist/esm/engine/organizations/resources/index.d.ts +1 -2
- package/dist/esm/engine/organizations/resources/index.d.ts.map +1 -1
- package/dist/esm/engine/organizations/resources/index.js +1 -2
- package/dist/esm/engine/organizations/resources/index.js.map +1 -1
- package/dist/esm/engine/organizations/resources/organization-members.resource.specification.d.ts +5 -4
- package/dist/esm/engine/organizations/resources/organization-members.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/organizations/resources/organization-members.resource.specification.js +114 -51
- package/dist/esm/engine/organizations/resources/organization-members.resource.specification.js.map +1 -1
- package/dist/esm/engine/organizations/resources/organization-profiles.resource.specification.d.ts +11 -10
- package/dist/esm/engine/organizations/resources/organization-profiles.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/organizations/resources/organization-profiles.resource.specification.js +10 -7
- package/dist/esm/engine/organizations/resources/organization-profiles.resource.specification.js.map +1 -1
- package/dist/esm/engine/organizations/resources/{service-provider-relationships.resource.specification.d.ts → organization-relationships.resource.specification.d.ts} +73 -56
- package/dist/esm/engine/organizations/resources/organization-relationships.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/organizations/resources/organization-relationships.resource.specification.js +643 -0
- package/dist/esm/engine/organizations/resources/organization-relationships.resource.specification.js.map +1 -0
- package/dist/esm/engine/organizations/resources/organization-unit-members.resource.specification.d.ts +1 -1
- package/dist/esm/engine/organizations/resources/organization-unit-members.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/organizations/resources/organization-unit-members.resource.specification.js +6 -4
- package/dist/esm/engine/organizations/resources/organization-unit-members.resource.specification.js.map +1 -1
- package/dist/esm/engine/organizations/resources/organization-units.resource.specification.js +2 -2
- package/dist/esm/engine/organizations/resources/organization-units.resource.specification.js.map +1 -1
- package/dist/esm/engine/organizations/resources/organizations.resource.specification.d.ts +4 -4
- package/dist/esm/engine/organizations/resources/organizations.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/organizations/resources/organizations.resource.specification.js +93 -43
- package/dist/esm/engine/organizations/resources/organizations.resource.specification.js.map +1 -1
- package/dist/esm/engine/organizations/resources/professional-qualifications.resource.specification.d.ts +39 -0
- package/dist/esm/engine/organizations/resources/professional-qualifications.resource.specification.d.ts.map +1 -0
- package/dist/esm/engine/organizations/resources/professional-qualifications.resource.specification.js +140 -0
- package/dist/esm/engine/organizations/resources/professional-qualifications.resource.specification.js.map +1 -0
- package/dist/esm/engine/providers/github.provider.specification.d.ts.map +1 -1
- package/dist/esm/engine/providers/github.provider.specification.js +8 -5
- package/dist/esm/engine/providers/github.provider.specification.js.map +1 -1
- package/dist/esm/engine/providers/google-drive.provider.specification.d.ts +6 -0
- package/dist/esm/engine/providers/google-drive.provider.specification.d.ts.map +1 -0
- package/dist/esm/engine/providers/google-drive.provider.specification.js +39 -0
- package/dist/esm/engine/providers/google-drive.provider.specification.js.map +1 -0
- package/dist/esm/engine/providers/index.d.ts +2 -0
- package/dist/esm/engine/providers/index.d.ts.map +1 -1
- package/dist/esm/engine/providers/index.js +5 -1
- package/dist/esm/engine/providers/index.js.map +1 -1
- package/dist/esm/engine/providers/microsoft-onedrive.provider.specification.d.ts +6 -0
- package/dist/esm/engine/providers/microsoft-onedrive.provider.specification.d.ts.map +1 -0
- package/dist/esm/engine/providers/microsoft-onedrive.provider.specification.js +39 -0
- package/dist/esm/engine/providers/microsoft-onedrive.provider.specification.js.map +1 -0
- package/dist/esm/engine/providers/mistral.provider.specification.js +2 -2
- package/dist/esm/engine/providers/mistral.provider.specification.js.map +1 -1
- package/dist/esm/engine/providers/resend.provider.specification.d.ts.map +1 -1
- package/dist/esm/engine/providers/resend.provider.specification.js +8 -6
- package/dist/esm/engine/providers/resend.provider.specification.js.map +1 -1
- package/dist/esm/engine/roles/app-audit-evidence-reader.role.specification.js +1 -1
- package/dist/esm/engine/roles/app-audit-evidence-reader.role.specification.js.map +1 -1
- package/dist/esm/engine/roles/index.d.ts +2 -2
- package/dist/esm/engine/roles/index.d.ts.map +1 -1
- package/dist/esm/engine/roles/index.js +5 -5
- package/dist/esm/engine/roles/index.js.map +1 -1
- package/dist/esm/engine/roles/org-delegated-capabilities.role.specification.d.ts +1 -1
- package/dist/esm/engine/roles/org-delegated-capabilities.role.specification.d.ts.map +1 -1
- package/dist/esm/engine/roles/org-delegated-capabilities.role.specification.js +24 -24
- package/dist/esm/engine/roles/org-delegated-capabilities.role.specification.js.map +1 -1
- package/dist/esm/engine/roles/org-representative.role.specification.d.ts +2 -0
- package/dist/esm/engine/roles/org-representative.role.specification.d.ts.map +1 -0
- package/dist/esm/engine/roles/org-representative.role.specification.js +22 -0
- package/dist/esm/engine/roles/org-representative.role.specification.js.map +1 -0
- package/dist/esm/engine/users/resources/anonymous-users.resource.specification.d.ts +1 -0
- package/dist/esm/engine/users/resources/anonymous-users.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/users/resources/anonymous-users.resource.specification.js +7 -5
- package/dist/esm/engine/users/resources/anonymous-users.resource.specification.js.map +1 -1
- package/dist/esm/engine/users/resources/inbound-contacts.resource.specification.d.ts +3 -3
- package/dist/esm/engine/users/resources/inbound-contacts.resource.specification.js +3 -3
- package/dist/esm/engine/users/resources/inbound-contacts.resource.specification.js.map +1 -1
- package/dist/esm/engine/users/resources/user-credentials.resource.specification.js +3 -3
- package/dist/esm/engine/users/resources/user-credentials.resource.specification.js.map +1 -1
- package/dist/esm/engine/users/resources/user-identity-links.resource.specification.d.ts +1 -1
- package/dist/esm/engine/users/resources/user-identity-links.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/users/resources/user-identity-links.resource.specification.js +13 -1
- package/dist/esm/engine/users/resources/user-identity-links.resource.specification.js.map +1 -1
- package/dist/esm/engine/users/resources/user-preferences-api-reference.examples.d.ts +1 -0
- package/dist/esm/engine/users/resources/user-preferences-api-reference.examples.d.ts.map +1 -1
- package/dist/esm/engine/users/resources/user-preferences-api-reference.examples.js +2 -0
- package/dist/esm/engine/users/resources/user-preferences-api-reference.examples.js.map +1 -1
- package/dist/esm/engine/users/resources/user-preferences.resource.specification.d.ts +4 -0
- package/dist/esm/engine/users/resources/user-preferences.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/users/resources/user-preferences.resource.specification.js +3 -3
- package/dist/esm/engine/users/resources/user-preferences.resource.specification.js.map +1 -1
- package/dist/esm/engine/users/resources/user-profiles.resource.specification.js +1 -1
- package/dist/esm/engine/users/resources/user-profiles.resource.specification.js.map +1 -1
- package/dist/esm/engine/users/resources/user-self-preferences.resource.specification.d.ts +4 -0
- package/dist/esm/engine/users/resources/user-self-preferences.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/users/resources/user-self-preferences.resource.specification.js +6 -2
- package/dist/esm/engine/users/resources/user-self-preferences.resource.specification.js.map +1 -1
- package/dist/esm/engine/users/resources/user-self.resource.specification.d.ts +1 -0
- package/dist/esm/engine/users/resources/user-self.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/users/resources/user-self.resource.specification.js +33 -0
- package/dist/esm/engine/users/resources/user-self.resource.specification.js.map +1 -1
- package/dist/esm/engine/users/resources/users.resource.specification.d.ts +2 -1
- package/dist/esm/engine/users/resources/users.resource.specification.d.ts.map +1 -1
- package/dist/esm/engine/users/resources/users.resource.specification.js +10 -4
- package/dist/esm/engine/users/resources/users.resource.specification.js.map +1 -1
- package/dist/esm/technical-documentation/engine-documentation-specification-catalog.companion.d.ts.map +1 -1
- package/dist/esm/technical-documentation/engine-documentation-specification-catalog.companion.js +17 -15
- package/dist/esm/technical-documentation/engine-documentation-specification-catalog.companion.js.map +1 -1
- package/package.json +8 -8
- package/dist/esm/engine/notifications/system-emails/subscription.system-emails.specification.d.ts +0 -6
- package/dist/esm/engine/notifications/system-emails/subscription.system-emails.specification.d.ts.map +0 -1
- package/dist/esm/engine/notifications/system-emails/subscription.system-emails.specification.js +0 -146
- package/dist/esm/engine/notifications/system-emails/subscription.system-emails.specification.js.map +0 -1
- package/dist/esm/engine/organizations/resources/service-provider-activity-feed.resource.specification.d.ts +0 -38
- package/dist/esm/engine/organizations/resources/service-provider-activity-feed.resource.specification.d.ts.map +0 -1
- package/dist/esm/engine/organizations/resources/service-provider-activity-feed.resource.specification.js +0 -152
- package/dist/esm/engine/organizations/resources/service-provider-activity-feed.resource.specification.js.map +0 -1
- package/dist/esm/engine/organizations/resources/service-provider-relationships.resource.specification.d.ts.map +0 -1
- package/dist/esm/engine/organizations/resources/service-provider-relationships.resource.specification.js +0 -417
- package/dist/esm/engine/organizations/resources/service-provider-relationships.resource.specification.js.map +0 -1
- package/dist/esm/engine/roles/org-service-provider.role.specification.d.ts +0 -2
- package/dist/esm/engine/roles/org-service-provider.role.specification.d.ts.map +0 -1
- package/dist/esm/engine/roles/org-service-provider.role.specification.js +0 -22
- package/dist/esm/engine/roles/org-service-provider.role.specification.js.map +0 -1
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { CoreAuditableEventType, EXTERNAL_DATA_PIPELINE_RUN_OPERATION_IDENTIFIER, OrganizationMembers_Operations, PlatformAccessGrantLifecycleTransition, ProviderCredentialDeletionReason, ResourceOperationDenialReason, } from "@wildo-ai/saas-models";
|
|
1
|
+
import { CoreAuditableEventType, EXTERNAL_DATA_PIPELINE_RUN_OPERATION_IDENTIFIER, HumanDecision_RefusalReason, OrganizationMembers_Operations, PlatformAccessGrantLifecycleTransition, ProviderCredentialDeletionReason, ResourceOperationDenialReason, } from "@wildo-ai/saas-models";
|
|
2
2
|
import { AssuranceEvidenceExecutionPlane, AssuranceEvidenceHome, } from "./assurance-program.schemas.js";
|
|
3
3
|
/**
|
|
4
4
|
* What each `evidence-requirement:*` slug actually ASKS FOR.
|
|
@@ -342,6 +342,110 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
342
342
|
+ "host-retained and evidenceable only as long as the host keeps its logs; C3's records are kept for the AUDIT_LOGS "
|
|
343
343
|
+ "period.",
|
|
344
344
|
},
|
|
345
|
+
/*
|
|
346
|
+
* #1845: the legal-hold half of WILDO.PRIVACY.STORAGE_LIMITATION. A hold is the EXCEPTION to a retention period, so it
|
|
347
|
+
* is a demand of its own beside `retention-window-declaration` rather than a clause of it: that demand asks whether the
|
|
348
|
+
* declared periods are enforced, this one asks whether a hold placed against them was recorded and honoured. Every
|
|
349
|
+
* clause reads what the mechanism already writes; nothing here asks the application to record anything.
|
|
350
|
+
*/
|
|
351
|
+
"evidence-requirement:legal-hold-record": {
|
|
352
|
+
schemaVersion: 1,
|
|
353
|
+
applicability: "APPLICABLE to every application: capability:legal-hold and capability:audit-events are ALWAYS_AVAILABLE. C1 has rows "
|
|
354
|
+
+ "only where someone placed or released a hold in the window, and an application that never placed one has failed "
|
|
355
|
+
+ "nothing: a judge records the quiet window as such. C2, C3 and C4 apply where their deletion path ran in the window "
|
|
356
|
+
+ "(a retention-class sweep over a bound resource, a tenant purge with a due tenant, a subject erasure); a path that "
|
|
357
|
+
+ "did not run has nothing to report about holds.",
|
|
358
|
+
clauses: [
|
|
359
|
+
{
|
|
360
|
+
clauseRef: "C1",
|
|
361
|
+
obligation: "Placing and releasing a legal hold is recorded on the held organization's audit trail: who did it, when, which "
|
|
362
|
+
+ "scope (the tenant, rows of one resource, or one data subject), which ground and which tier, and never the "
|
|
363
|
+
+ "free-text matter.",
|
|
364
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
365
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
366
|
+
homeRationale: "`AuditLogsBackendService.logLegalHoldLifecycle` writes one HIGH row per transition, after the hold's own write "
|
|
367
|
+
+ "commits, on the HELD tenant's trail whichever tier placed it, so the tenant reads every hold on its data in its "
|
|
368
|
+
+ "own audit logs. The row carries the hold id, the held organization, `scopeKind`, `legalGround`, `placementTier`, "
|
|
369
|
+
+ "the held resource type and row count or the subject, and `placedByUserId` or `releasedByUserId` (a release also "
|
|
370
|
+
+ "carries `heldForSeconds`). The matter text the hold stores is never copied into it. The write is best-effort and "
|
|
371
|
+
+ "the hold stands either way, so a hold with no row is a LOSS, not a hold that was never placed, and a loss is "
|
|
372
|
+
+ "recorded in one of two places: an emission that never reached the audit service (unbound, or it threw) is a "
|
|
373
|
+
+ "dropped `LEGAL_HOLD_LIFECYCLE` observation (`audit_observation_dropped_total`), while a row the audit store "
|
|
374
|
+
+ "refused to write is logged and counted as a store failure (`audit_store_failure_total`), because the default "
|
|
375
|
+
+ "REPORT_AND_CONTINUE policy keeps the hold's request from failing. Look for a missing row in both. The collector reports how "
|
|
376
|
+
+ "the rows divide over `legalGround`; every ground is evidence, so none is filtered out.",
|
|
377
|
+
emittedCountLimit: {
|
|
378
|
+
unreached: "each row names the hold's scope, ground, tier and actor, never whether the hold was justified or that it covered every row the matter needed",
|
|
379
|
+
},
|
|
380
|
+
requiredAuditEventMembers: [CoreAuditableEventType.LEGAL_HOLD_PLACED, CoreAuditableEventType.LEGAL_HOLD_RELEASED],
|
|
381
|
+
reportedAuditEventDataDistribution: { field: "legalGround" },
|
|
382
|
+
},
|
|
383
|
+
{
|
|
384
|
+
clauseRef: "C2",
|
|
385
|
+
obligation: "The retention-class sweep deletes no row a hold covers: it counts the rows past their period that a hold kept, and "
|
|
386
|
+
+ "when the holds cannot be read it deletes nothing more for that resource and records the run as failed.",
|
|
387
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
388
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
389
|
+
homeRationale: "The same ENGINE-owned store as `retention-window-declaration` C3, `retentionEnforcementRecords`, read through "
|
|
390
|
+
+ "`GET /assurance/retention-enforcement-evidence`: each per-resource record carries `heldSkippedCount`, the rows "
|
|
391
|
+
+ "past their period the run left because a hold covers them, and a run whose hold read failed is a FAILED record "
|
|
392
|
+
+ "with `failureStage: legal-hold-unverifiable`. An ABSENT count means no hold check ran for that resource on that "
|
|
393
|
+
+ "run, which is \"not measured\", never zero. A row held across several runs is counted by each, so the count is "
|
|
394
|
+
+ "read per resource from its latest run, never summed over the window.",
|
|
395
|
+
},
|
|
396
|
+
{
|
|
397
|
+
clauseRef: "C3",
|
|
398
|
+
obligation: "The tenant purge destroys no tenant under an active hold, any hold on it: the tenant stays marked and is counted, "
|
|
399
|
+
+ "and when the holds cannot be read no tenant is purged in that run.",
|
|
400
|
+
home: AssuranceEvidenceHome.PERFORMED,
|
|
401
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
402
|
+
homeRationale: "The organization-purge batch returns `heldByLegalHold` and `legalHoldStateUnverifiable` and logs the first on its "
|
|
403
|
+
+ "`Organization purge batch completed` line; a run that could not read the holds logs an error and purges "
|
|
404
|
+
+ "nothing. No engine store keeps either, and job-execution output expires in 24 hours, so the evidence is the "
|
|
405
|
+
+ "runtime's log in the accountable host's archive, which is why this is PERFORMED. Each held tenant is also "
|
|
406
|
+
+ "logged with its hold ids and grounds.",
|
|
407
|
+
},
|
|
408
|
+
{
|
|
409
|
+
clauseRef: "C4",
|
|
410
|
+
obligation: "A subject erasure leaves the rows a hold covers as they are and reports them with their grounds, instead of "
|
|
411
|
+
+ "erasing them or refusing the whole erasure.",
|
|
412
|
+
home: AssuranceEvidenceHome.PERFORMED,
|
|
413
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
414
|
+
homeRationale: "The erasure report's `heldByLegalHold` lists each kept row with its resource type and grounds; it is the "
|
|
415
|
+
+ "erasure's RETURN VALUE and the engine persists none of it (the application keeps its receipt, as "
|
|
416
|
+
+ "`erasure-propagation-record` C1 describes). What the engine itself writes is the count, `legallyHeldRowCount`, "
|
|
417
|
+
+ "on its `Subject erasure completed` log line, so the engine's own evidence is that line in the host's archive. "
|
|
418
|
+
+ "The line is written inside the erasure's transaction, before it commits: a commit that fails after it leaves a "
|
|
419
|
+
+ "`completed` line for an erasure that rolled back, so read the line beside the erasure's outcome, never alone. "
|
|
420
|
+
+ "Which rows and which grounds are read from the application's retained receipts where it keeps them. The line is "
|
|
421
|
+
+ "read from the BACKEND runtime's log: an erasure a minion or worker runtime performs through its own system "
|
|
422
|
+
+ "access logs to that runtime and is not counted here.",
|
|
423
|
+
},
|
|
424
|
+
],
|
|
425
|
+
minimumEvidence: "C1: the window's LEGAL_HOLD_PLACED and LEGAL_HOLD_RELEASED rows, counted per member, with the distribution of their "
|
|
426
|
+
+ "legalGround. C2: per bound resource, the latest retention-enforcement record's heldSkippedCount, and every record "
|
|
427
|
+
+ "failed at legal-hold-unverifiable. C3: the tenant purge's completed runs in the window from the host's log archive, "
|
|
428
|
+
+ "with heldByLegalHold summed and every run that could not read the holds. C4: the subject erasures completed in the "
|
|
429
|
+
+ "window from the same archive, with legallyHeldRowCount summed.",
|
|
430
|
+
verificationSteps: [
|
|
431
|
+
"Record the window, the application revision and the runtime profile.",
|
|
432
|
+
"Count the LEGAL_HOLD_PLACED and LEGAL_HOLD_RELEASED rows (C1) and read their grounds. A hold placed before the window and released in it shows only its release row.",
|
|
433
|
+
"Read the retention-enforcement records (C2) per bound resource, from each one's latest run: heldSkippedCount, and any record failed at legal-hold-unverifiable.",
|
|
434
|
+
"Read the tenant purge's completed runs (C3) and sum heldByLegalHold; name every run whose holds could not be read.",
|
|
435
|
+
"Read the completed subject erasures (C4) and sum legallyHeldRowCount; where the application retains its erasure receipts, read which rows and grounds they list.",
|
|
436
|
+
"Set C2, C3 and C4 beside C1: rows kept by a hold while no hold is recorded as placed, in the window or before it, is a finding to explain.",
|
|
437
|
+
],
|
|
438
|
+
severityGuidance: "A row, tenant or erasure kept by a hold is the control WORKING, never a retention failure. A run that could not read "
|
|
439
|
+
+ "the holds and therefore deleted nothing is the control failing CLOSED, also not a nonconformity; repeated runs of it "
|
|
440
|
+
+ "are an availability finding, because retention stops for as long as the holds stay unreadable. The nonconformity is "
|
|
441
|
+
+ "a held row a deletion path destroyed, or a hold placed or released with no row recording it.",
|
|
442
|
+
limits: "It evidences that holds were recorded and that the deletion paths reported what they kept, never that a hold was "
|
|
443
|
+
+ "justified, covered what the matter required, or was released when the matter ended: a hold has no expiry, by design. "
|
|
444
|
+
+ "C1 and C2 read engine stores kept for the AUDIT_LOGS period; C3 and C4 are host-retained and evidenceable only as "
|
|
445
|
+
+ "long as the host keeps its logs. The interactive deletes, scrubs, cascades and the file cleanup refuse or skip a "
|
|
446
|
+
+ "held row too, and leave no count this demand reads: an interactive refusal answers 409 `legal_hold_active` to the "
|
|
447
|
+
+ "caller, and the file cleanup logs its own completion line.",
|
|
448
|
+
},
|
|
345
449
|
"evidence-requirement:privileged-access-review-occurrence": {
|
|
346
450
|
schemaVersion: 1,
|
|
347
451
|
applicability: "APPLICABLE to every application: both required capabilities are ALWAYS_AVAILABLE — capability:audit-events is the "
|
|
@@ -423,8 +527,8 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
423
527
|
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
424
528
|
homeRationale: "The application's own resource row is the durable record, and no CoreAuditableEventType member covers the "
|
|
425
529
|
+ "request LIFECYCLE — intake, withdrawal, execution. The erasure-shaped members that do exist record what an "
|
|
426
|
-
+ "erasure did to a particular STORE (
|
|
427
|
-
+ "provider credential), which is a different question and is demanded by "
|
|
530
|
+
+ "erasure did to a particular STORE (rows a forwarded erasure deleted in a system of record, a recipient's "
|
|
531
|
+
+ "erasure outcome, a person's provider credential), which is a different question and is demanded by "
|
|
428
532
|
+ "`evidence-requirement:erasure-propagation-record`, not here. The row's schema and status vocabulary are the "
|
|
429
533
|
+ "application's, so a collector queries it against that application's specification.",
|
|
430
534
|
},
|
|
@@ -509,8 +613,11 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
509
613
|
+ "call with no route, so an application that never erases a subject has produced no accounting, no record and has "
|
|
510
614
|
+ "failed nothing — a judge records not_applicable_as_configured for those three clauses. C2 applies to every "
|
|
511
615
|
+ "application: it is the framework's own behaviour and is driven, not gathered. C5 applies where the application "
|
|
512
|
-
+ "declares an externalDataPipeline — a
|
|
513
|
-
+ "
|
|
616
|
+
+ "declares an INBOUND externalDataPipeline with the OWNED_COPY posture — a copy of a system of record's records the "
|
|
617
|
+
+ "application keeps beyond its source (#2560). A following copy (the default) is not C5's: the erasure leaves it to its "
|
|
618
|
+
+ "source, which answers requests about its own records, and a record the source removes is removed or scrubbed in the "
|
|
619
|
+
+ "copy at the next complete pass, as its orphan policy declares, so no suppression is owed. An application that "
|
|
620
|
+
+ "declares no owned copy has no C5 to evidence, however many erasures it runs.",
|
|
514
621
|
clauses: [
|
|
515
622
|
{
|
|
516
623
|
clauseRef: "C1",
|
|
@@ -547,21 +654,22 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
547
654
|
},
|
|
548
655
|
{
|
|
549
656
|
clauseRef: "C3",
|
|
550
|
-
obligation: "Where erasure
|
|
551
|
-
+ "
|
|
657
|
+
obligation: "Where erasure is forwarded to a system of record and deletes rows there, it leaves a durable record naming the "
|
|
658
|
+
+ "subject, the resource and the external system.",
|
|
552
659
|
home: AssuranceEvidenceHome.EMITTED,
|
|
553
660
|
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
554
|
-
homeRationale: "
|
|
555
|
-
+ "this control and from no other
|
|
556
|
-
+ "
|
|
557
|
-
+ "
|
|
558
|
-
+ "
|
|
559
|
-
+ "
|
|
560
|
-
+ "
|
|
661
|
+
homeRationale: "The row is emitted by the per-resource external effect the erasure runs post-commit, which is reached from "
|
|
662
|
+
+ "this control and from no other, and carries the remote row count. Only a binding that FORWARDS erasures writes "
|
|
663
|
+
+ "it: a system of record that handles requests about its own records (the default since #2558) is named on C1's "
|
|
664
|
+
+ "receipt as handled at the source, and there was no CRITICAL acknowledgement row to count after #2559 removed the "
|
|
665
|
+
+ "per-request refusal it recorded. An effect that failed before sending anything writes no row. One that sent "
|
|
666
|
+
+ "requests writes its row even when it deleted nothing, marked `failed` when a step failed, so the requests it "
|
|
667
|
+
+ "sent are on the trail (#1970). No row, `failed` or not, proves the subject is fully erased at the source: "
|
|
668
|
+
+ "which steps failed is on C1's receipt, which is why C1 and C3 are read together. A billing-provider erasure "
|
|
669
|
+
+ "writes no row at all and is evidenced by C1 alone.",
|
|
561
670
|
emittedCountLimit: { unreached: "each row names the subject, the resource and the external system it reached" },
|
|
562
671
|
requiredAuditEventMembers: [
|
|
563
672
|
CoreAuditableEventType.HTTP_API_EXTERNAL_ERASURE,
|
|
564
|
-
CoreAuditableEventType.HTTP_API_NON_ERASURE_ACKNOWLEDGED,
|
|
565
673
|
],
|
|
566
674
|
},
|
|
567
675
|
{
|
|
@@ -594,19 +702,21 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
594
702
|
},
|
|
595
703
|
{
|
|
596
704
|
clauseRef: "C5",
|
|
597
|
-
obligation: "Where an erased row had been copied in from a system of record by an external-data pipeline
|
|
598
|
-
+ "suppresses that remote record, and no later pass — full or incremental — copies it back in; each pipeline run "
|
|
705
|
+
obligation: "Where an erased row had been copied in from a system of record by an external-data pipeline that keeps its copy "
|
|
706
|
+
+ "(OWNED_COPY), the erasure suppresses that remote record, and no later pass — full or incremental — copies it back in; each pipeline run "
|
|
599
707
|
+ "counts the rows it declined to load on its own audit row.",
|
|
600
708
|
home: AssuranceEvidenceHome.EMITTED,
|
|
601
709
|
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
602
710
|
homeRationale: "The source still holds a record the application erased locally, so without this the next scheduled pass writes "
|
|
603
711
|
+ "the person back while the erasure receipt says complete. The erasure writes a suppression at the two places every "
|
|
604
|
-
+ "erasure of
|
|
712
|
+
+ "erasure of an owned-copy row passes through — the scrub that retains a row, and the erasure cascade's delete — keyed by "
|
|
605
713
|
+ "a keyed hash of the system, the remote entity and the remote key, never the key itself. Both pipeline tracks skip "
|
|
606
714
|
+ "a suppressed key, and each run's read-through audit row carries `suppressedRows`, which the collector SUMS over "
|
|
607
715
|
+ "the window's pipeline runs. The row count says the runs happened; the sum says suppression held on them. A "
|
|
608
716
|
+ "suppression is released only by a completed full pass over the whole remote entity that no longer sees the "
|
|
609
|
-
+ "record, and a run that finds suppressions hashed under a key the application no longer derives refuses to load."
|
|
717
|
+
+ "record, and a run that finds suppressions hashed under a key the application no longer derives refuses to load. "
|
|
718
|
+
+ "Only an erasure writes one: a parent delete that scrubs an owned-copy row is not a subject's request and leaves "
|
|
719
|
+
+ "the next pass free to reload it, and a following copy writes none because its source decides.",
|
|
610
720
|
emittedCountLimit: {
|
|
611
721
|
unreached: "each run row says how many rows it declined to load, never which erasure each skip honoured, nor that every erased pipeline row was suppressed",
|
|
612
722
|
},
|
|
@@ -618,9 +728,12 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
618
728
|
},
|
|
619
729
|
{
|
|
620
730
|
clauseRef: "C6",
|
|
621
|
-
obligation: "Each erasure accounts for every
|
|
731
|
+
obligation: "Each erasure accounts for every recipient the application declares as holding the subject's data: one an effect "
|
|
622
732
|
+ "reached reports that effect's outcome, and one no effect reached reports the status its declared erasure stance "
|
|
623
|
-
+ "decides — unmeasured when it declares none — so no recipient passes in silence.
|
|
733
|
+
+ "decides — unmeasured when it declares none — so no recipient passes in silence. A vendor-hosted system of record "
|
|
734
|
+
+ "is one of them and declares no stance: it reports its HTTP_API resources' effects, or HANDLED_AT_SOURCE when "
|
|
735
|
+
+ "none applies to the subject (#2585). A forwarded erasure is also C3's, and each bound resource is named on C1's "
|
|
736
|
+
+ "receipt. A self-hosted system of record is the operator's own system, not a recipient.",
|
|
624
737
|
home: AssuranceEvidenceHome.EMITTED,
|
|
625
738
|
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
626
739
|
homeRationale: "GDPR Art. 19 asks the controller to communicate an erasure to each recipient the data was disclosed to, and to name "
|
|
@@ -631,6 +744,8 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
631
744
|
+ "whose account reached it, who owes the communication (the operator as controller, the operator assisting a "
|
|
632
745
|
+ "controller tenant, or the tenant for its own vendor) and the status. The collector reports how the rows divide "
|
|
633
746
|
+ "over `status`: an UNMEASURED or NOT_REGISTERED share is a vendor erasure does not reach, stated rather than hidden. "
|
|
747
|
+
+ "A COMPLETES_AT_VENDOR row is a vendor that deletes on its own published schedule, and carries the declared window "
|
|
748
|
+
+ "and the date it lapses, so the share that completes later is stated with when. "
|
|
634
749
|
+ "An application that declares no recipients leaves no row and marks every receipt with one unmeasured outcome, "
|
|
635
750
|
+ "which C1 reads.",
|
|
636
751
|
emittedCountLimit: {
|
|
@@ -646,22 +761,22 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
646
761
|
+ "and each external effect's ref and status. C2: a driven erasure showing a declined executing edge reported as not "
|
|
647
762
|
+ "executed, at the direct level and one level below a deleted dependent, the latter also turning its direct edge not "
|
|
648
763
|
+ "executed; an unclassified or deferred edge, at the direct level and below a deleted dependent, refusing before any "
|
|
649
|
-
+ "write; and each incomplete external status classified incomplete. C3: the external-erasure
|
|
650
|
-
+ "
|
|
764
|
+
+ "write; and each incomplete external status classified incomplete. C3: the external-erasure rows "
|
|
765
|
+
+ "for the window. C4: the credential-deletion rows carrying the erasure reason for the "
|
|
651
766
|
+ "window, one per credential each erasure destroyed, with how they divide over the vendor-revocation outcome. C5: the "
|
|
652
767
|
+ "pipeline-run read-through rows for the window with the sum of `suppressedRows` over them, and — once, where the "
|
|
653
|
-
+ "application
|
|
768
|
+
+ "application keeps an owned copy — a driven erasure of an owned-copy row followed by a full and an incremental pass "
|
|
654
769
|
+ "that load nothing under its record. Any run row carrying a non-zero `liveSuppressedRows` is listed. C6: the "
|
|
655
770
|
+ "recipient-outcome rows for the window, with how they divide over status, and the application's generated recipient "
|
|
656
771
|
+ "declaration they were selected from.",
|
|
657
772
|
verificationSteps: [
|
|
658
773
|
"Resolve applicability for C1, C3 and C4: does the application invoke subject erasure? If not, record those three clauses not_applicable_as_configured.",
|
|
659
774
|
"For C1, gather the application's retained receipts and assert each records, per erasure, the subject's impersonalization, the executed state of every executing direct edge, and every external effect's status. Do NOT assert a per-edge list or a descendant list: whether the application keeps edges individually or as a count is its own modelling choice.",
|
|
660
|
-
"Cross-check C1 against C3: for every receipt naming an external-system effect that deleted rows
|
|
775
|
+
"Cross-check C1 against C3: for every receipt naming an external-system effect that deleted rows, find the matching row for that subject and resource. A receipt with no row is reported; a row with no receipt is reported.",
|
|
661
776
|
"For C2, drive an erasure over a direct dependent edge whose teardown the cascade declines and assert the receipt says not executed; drive one whose direct edge runs but whose descendant is declined and assert both the descendant and its direct edge say not executed; drive one over an unclassified edge, directly and below a deleted dependent, and assert each refuses with nothing written; and assert every incomplete external status, including an acknowledged non-erasure, is classified incomplete.",
|
|
662
777
|
"For C4, cross-check each erasure whose credential edge executed against the credential-deletion rows carrying the erasure reason for that subject. A subject who held no credentials owes no row; a destroyed credential with no row is reported.",
|
|
663
778
|
"For C4, read the vendor-revocation distribution of those rows, and cross-check each erasure's receipt: its connected-account vendor-grant effect must name the same credentials with the same outcomes. A row carrying no vendor-revocation outcome predates the revocation and is reported as unrevoked, never as revoked.",
|
|
664
|
-
"For C5, resolve applicability: does the application declare an externalDataPipeline? Then read the window's pipeline-run rows and the sum of their suppressedRows. A sum of zero is not a finding on its own — no erased record may have been read in the window — but after an erasure of
|
|
779
|
+
"For C5, resolve applicability: does the application declare an INBOUND externalDataPipeline with the OWNED_COPY posture? A following copy is not C5's: record not_applicable_as_configured for an application whose copies all follow their source. Then read the window's pipeline-run rows and the sum of their suppressedRows. A sum of zero is not a finding on its own — no erased record may have been read in the window — but after an erasure of an owned-copy row, drive or observe the next full and incremental pass and assert that the erased record's row did not come back as a live row and that the run's suppressedRows counted it.",
|
|
665
780
|
"For C6, read the generated recipient declaration and confirm it is current (`wildo config subject-data-recipients --check`), then read the window's recipient-outcome rows and their status distribution. Cross-check C1: every receipt's `recipient:<ref>` external effects must match the rows for that erasure. A declared operator-account recipient with no row for an erasure is reported; an UNMEASURED share names vendors whose erasure stance nobody declared.",
|
|
666
781
|
],
|
|
667
782
|
severityGuidance: "A receipt reporting an executing edge as run when its teardown was declined — or a direct edge as run when a "
|
|
@@ -676,7 +791,7 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
676
791
|
+ "credential is destroyed and the outcome is recorded, and what is owed is a sentence to the person telling them to "
|
|
677
792
|
+ "remove the application in the vendor's own settings. A receipt reporting that vendor effect as complete while a "
|
|
678
793
|
+ "row says a grant was left live is a MAJOR, for the same reason an incomplete external status read as complete is. "
|
|
679
|
-
+ "An erased
|
|
794
|
+
+ "An erased owned-copy row that a later pass loaded again as a live row is a MAJOR against the framework: the "
|
|
680
795
|
+ "erasure was silently undone; a run row reporting `liveSuppressedRows` above zero is exactly that finding, "
|
|
681
796
|
+ "surfaced. A pipeline run refused because its suppressions were hashed under a rotated key is an "
|
|
682
797
|
+ "OBSERVATION against the application's key handling, not a nonconformity: the refusal is the control holding. "
|
|
@@ -705,8 +820,12 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
705
820
|
+ "and never remove; it is not self-healing, only DETECTED — each run counts `liveSuppressedRows`, and a non-zero "
|
|
706
821
|
+ "count is an erasure that did not hold. Two pipelines keying one entity differently do not share a suppression. "
|
|
707
822
|
+ "C6's recipient list is only as complete as the generated declaration: a vendor added since it was generated is "
|
|
708
|
-
+ "not accounted for until it is regenerated.
|
|
709
|
-
+ "
|
|
823
|
+
+ "not accounted for until it is regenerated. A vendor-hosted system of record is in this list (#2585), and its outcome "
|
|
824
|
+
+ "is what its HTTP_API resources' effects reported, or HANDLED_AT_SOURCE when none applies to the subject: that says "
|
|
825
|
+
+ "the system answers for its own records, not that anything was erased there. Each bound resource is also on C1's "
|
|
826
|
+
+ "receipt under its binding's subject-rights posture, and a forwarded erasure is also C3's. A self-hosted system of "
|
|
827
|
+
+ "record is the operator's own system and is not a recipient. The generator does not read a system of record's "
|
|
828
|
+
+ "resource flows, so the list names it for every subject as one that may hold their records.",
|
|
710
829
|
},
|
|
711
830
|
"evidence-requirement:selected-processor-declaration": {
|
|
712
831
|
schemaVersion: 1,
|
|
@@ -741,7 +860,7 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
741
860
|
+ "DPA that governs it and the vendor's retention (`processorAgreementStanding`). "
|
|
742
861
|
+ "Per source of personal data obtained from a system of record rather than from the person (#1745): the resource bound to "
|
|
743
862
|
+ "or piped from it, the instance and its vendor, the categories that come from it, the purposes its binding links through "
|
|
744
|
-
+ "`purposeRefs`, and for a local copy whether its rows are kept until erased (`deriveSystemOfRecordSources` — the same "
|
|
863
|
+
+ "`purposeRefs`, where requests about its records are answered (its subject-rights posture, #2561), and for a local copy whether its rows are kept until erased (`deriveSystemOfRecordSources` — the same "
|
|
745
864
|
+ "derivation the privacy notice's `system-of-record-sources` fact states, GDPR Art. 14(2)(f)). "
|
|
746
865
|
+ "C2: that the extraction seam REFUSES when no provider is resolved rather than degrading, and that its one "
|
|
747
866
|
+ "functional caller serves the local library when the field did not opt in.",
|
|
@@ -760,7 +879,7 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
760
879
|
+ "never as an empty list, and a list including free text as a FLOOR. Name each recipient linked to no purpose, and each "
|
|
761
880
|
+ "`purposeRef` naming no declared lawful basis.",
|
|
762
881
|
"Derive, per resource bound to or piped from a system of record, the SOURCE the notice owes for data not collected from the "
|
|
763
|
-
+ "person (`deriveSystemOfRecordSources`, #1745): the instance, its vendor, the categories, the linked purposes, and for a copy "
|
|
882
|
+
+ "person (`deriveSystemOfRecordSources`, #1745): the instance, its vendor, the categories, the linked purposes, its subject-rights posture (#2561), and for a copy "
|
|
764
883
|
+ "whether it is kept until erased. Name each source linked to no purpose and each `purposeRef` naming no declared lawful basis; "
|
|
765
884
|
+ "report flows that could not be read as NOT DERIVED, never as no source.",
|
|
766
885
|
"Read the operator's contractual record per register entry (`processorAgreementStanding`, #1709): which a recorded DPA governs, "
|
|
@@ -806,6 +925,885 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
806
925
|
+ "instance holds those rows on the operator's account. The DPA status is the operator's statement; nothing here verifies "
|
|
807
926
|
+ "the contract exists.",
|
|
808
927
|
},
|
|
928
|
+
"evidence-requirement:agent-tool-authority-declaration": {
|
|
929
|
+
schemaVersion: 1,
|
|
930
|
+
applicability: "Applies to every application, and within it to every function tool it registers for its agents. An application that registers "
|
|
931
|
+
+ "none has an empty population, which is the answer rather than a gap. Tools a third-party MCP server serves run on that server's "
|
|
932
|
+
+ "authority and are outside the population.",
|
|
933
|
+
clauses: [
|
|
934
|
+
{
|
|
935
|
+
clauseRef: "C1",
|
|
936
|
+
obligation: "Every tool an agent may call declares how it reaches the application's data, and a tool that reaches beyond its caller's own "
|
|
937
|
+
+ "authority says why. A registered tool no agent names cannot be called, and is listed rather than graded.",
|
|
938
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
939
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
940
|
+
homeRationale: "The engine closes every tool over its caller's server-verified authority, and its own tool factories dispatch through the "
|
|
941
|
+
+ "operation's gates for that caller. What a hand-written tool's code then does is visible to no runtime check, so the tool "
|
|
942
|
+
+ "declares it (`FunctionTool.authority`). The committed declaration is the evidence of which tools stay within their caller's "
|
|
943
|
+
+ "authority, and of which ones say they do not.",
|
|
944
|
+
},
|
|
945
|
+
{
|
|
946
|
+
clauseRef: "C2",
|
|
947
|
+
obligation: "A tool call the calling principal's own authority does not admit is refused, and the refusal reaches the agent.",
|
|
948
|
+
home: AssuranceEvidenceHome.UNPRODUCED,
|
|
949
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
950
|
+
homeRationale: "The refusal is the authorizer's, recorded by its denial funnel for every plane alike, so a denial row does not say an AGENT "
|
|
951
|
+
+ "was refused. Showing it for an agent needs a driven turn whose tool the principal may not run, and no scenario drives one "
|
|
952
|
+
+ "yet (#2850). UNPRODUCED is the honest value until one does.",
|
|
953
|
+
},
|
|
954
|
+
],
|
|
955
|
+
minimumEvidence: "C1: the function tools the application registers, each with its declared authority kind and reason, whether it can change data, "
|
|
956
|
+
+ "whether a person approves it, and the agents that may call it.",
|
|
957
|
+
verificationSteps: [
|
|
958
|
+
"Read the agents and tools the application registers through its backend's initialization declaration.",
|
|
959
|
+
"For each tool, read its declared authority and check it against the declaration schema.",
|
|
960
|
+
"Name every tool that declares none, every declaration the schema refuses, and every tool declaring system authority with its reason.",
|
|
961
|
+
],
|
|
962
|
+
severityGuidance: "A tool that declares no authority is a MAJOR: the confinement claim cannot be made for it, and it may reach data as the system. A "
|
|
963
|
+
+ "declaration the schema refuses is the same finding. A tool that declares system authority with a reason is an OBSERVATION: it "
|
|
964
|
+
+ "is visible and argued, and the claim does not cover it.",
|
|
965
|
+
limits: "A declaration is a claim about the tool's code, not a proof of it: a tool that declares its caller's authority and writes through "
|
|
966
|
+
+ "a `*AsSystem` door is not caught here. Tools served by a third-party MCP server run on that server's authority and are not "
|
|
967
|
+
+ "covered.",
|
|
968
|
+
},
|
|
969
|
+
"evidence-requirement:agent-model-boundary-personal-data": {
|
|
970
|
+
schemaVersion: 1,
|
|
971
|
+
applicability: "Applies to every application with an agent on an AI capability of the contracted service. An application with none sends "
|
|
972
|
+
+ "no record to a model through an agent, which is the answer rather than a gap.",
|
|
973
|
+
clauses: [
|
|
974
|
+
{
|
|
975
|
+
clauseRef: "C1",
|
|
976
|
+
obligation: "Whether an agent's model is sent the values of personal fields, or placeholders in their place (removed or tokenized), is a "
|
|
977
|
+
+ "declared decision per agent.",
|
|
978
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
979
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
980
|
+
homeRationale: "The treatment is the agent's committed `contentGuardrails.personalData`, applied by the engine on every tool result the model "
|
|
981
|
+
+ "reads. Startup refuses a guarded agent offered a tool that does not say what its result carries (`personalDataOutput`), so a "
|
|
982
|
+
+ "declared treatment always applies. The commit is the evidence of what the agent's own declaration sends a model.",
|
|
983
|
+
},
|
|
984
|
+
{
|
|
985
|
+
clauseRef: "C2",
|
|
986
|
+
obligation: "A model guarded by a treatment never receives the value of a personal field, through any tool's result.",
|
|
987
|
+
home: AssuranceEvidenceHome.UNPRODUCED,
|
|
988
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
989
|
+
homeRationale: "What a model was sent is neither stored nor recorded, by design: storing it would be the disclosure the control prevents. "
|
|
990
|
+
+ "Showing it needs a driven turn against a model double that reports what it received, and no scenario drives one yet (#2850).",
|
|
991
|
+
},
|
|
992
|
+
],
|
|
993
|
+
minimumEvidence: "C1: per agent on a contracted-service AI capability, its declared personal-data treatment or its absence.",
|
|
994
|
+
verificationSteps: [
|
|
995
|
+
"Read the agents and their tools through the backend's initialization declaration.",
|
|
996
|
+
"Name every agent declaring no personal-data treatment of its own.",
|
|
997
|
+
],
|
|
998
|
+
severityGuidance: "An agent declaring no treatment of its own is a MINOR: its model receives personal values unless an organization's agent policy "
|
|
999
|
+
+ "imposes a treatment, which may be right, and nobody has stated the decision.",
|
|
1000
|
+
limits: "It evidences the agents' own declarations, not what a given model call received (C2), and not a treatment an organization's "
|
|
1001
|
+
+ "agent policy imposes at run time. A tool that declares its result carries no record data while returning some is not caught here.",
|
|
1002
|
+
},
|
|
1003
|
+
"evidence-requirement:agent-human-oversight-record": {
|
|
1004
|
+
schemaVersion: 1,
|
|
1005
|
+
applicability: "Applies to every application with an AI use in the contracted service, and within it to every use declared to affect a person. "
|
|
1006
|
+
+ "An application whose agents raise no decision has no rows for C2, which is an empty population rather than a gap, and one whose "
|
|
1007
|
+
+ "administrators suspended nothing has no rows for C4, likewise. Both C4 rows are written best-effort, so a missing one is not "
|
|
1008
|
+
+ "read as a decision or a refusal that did not happen (C4's limit).",
|
|
1009
|
+
clauses: [
|
|
1010
|
+
{
|
|
1011
|
+
clauseRef: "C1",
|
|
1012
|
+
obligation: "Every AI use declared to affect a person reaches a person: its agent's tools that change data wait for an approval, or its "
|
|
1013
|
+
+ "declared human-review arrangement names how a person intervenes. A gate routed to a role or an assignee is decided by "
|
|
1014
|
+
+ "someone other than the person whose turn raised it.",
|
|
1015
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1016
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1017
|
+
homeRationale: "Which tools wait for a person, and who that person is, are committed declarations (`humanApproval`, `requiresHumanApproval`), "
|
|
1018
|
+
+ "and so is each use's consequence and stated review (`aiUse`). Joining them shows which declared reviews the gates evidence "
|
|
1019
|
+
+ "and which are stated only.",
|
|
1020
|
+
},
|
|
1021
|
+
{
|
|
1022
|
+
clauseRef: "C2",
|
|
1023
|
+
obligation: "Each held decision is recorded when it is raised and when it ends, and a deadline that passes unanswered is recorded.",
|
|
1024
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1025
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1026
|
+
homeRationale: "A durable row class in the application's own audit store, one family for the conversational tool approval, the actor-system "
|
|
1027
|
+
+ "suspension and the operation-approval hold, keyed by one decision ref. The rows are counted by member; the same family also "
|
|
1028
|
+
+ "records decisions that are not an AI agent's (a representative's admission), so a count is an upper bound for agent decisions.",
|
|
1029
|
+
emittedCountLimit: { unreached: "a count shows decisions were recorded, not that every held action was: whether an action needed a decision is C1's question" },
|
|
1030
|
+
requiredAuditEventMembers: [
|
|
1031
|
+
CoreAuditableEventType.HUMAN_DECISION_REQUESTED,
|
|
1032
|
+
CoreAuditableEventType.HUMAN_DECISION_RESOLVED,
|
|
1033
|
+
CoreAuditableEventType.HUMAN_DECISION_DEADLINE_MISSED,
|
|
1034
|
+
],
|
|
1035
|
+
},
|
|
1036
|
+
{
|
|
1037
|
+
clauseRef: "C3",
|
|
1038
|
+
obligation: "A person can stop a running agent turn.",
|
|
1039
|
+
home: AssuranceEvidenceHome.UNPRODUCED,
|
|
1040
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1041
|
+
homeRationale: "The stop ends the turn through the runtime's own arms and writes no record of its own, so no artifact evidences that a turn "
|
|
1042
|
+
+ "was stopped. An administrator's suspension of an agent (#2843) is C4 and C5. UNPRODUCED is the honest value for this one.",
|
|
1043
|
+
},
|
|
1044
|
+
{
|
|
1045
|
+
clauseRef: "C4",
|
|
1046
|
+
obligation: "Each suspension and resumption of an agent, by an organization's administrator or application-wide by the super-administrator "
|
|
1047
|
+
+ "with a justification, is recorded with its scope, the agent, the reason and whether it changed anything; and each call refused "
|
|
1048
|
+
+ "because its agent is suspended, or because its suspension could not be read, is recorded with the agent, the reason, the "
|
|
1049
|
+
+ "scope that suspended it and the door, when the call was made in a conversation or a flow run.",
|
|
1050
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1051
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1052
|
+
homeRationale: "Two row classes in the application's own audit store: the administrator's decision (`AGENT_SUSPENSION_CHANGED`, written after "
|
|
1053
|
+
+ "the change commits, CRITICAL for an application-wide one, HIGH for an organization's) and the agent service's refusal "
|
|
1054
|
+
+ "(`SUSPENDED_AGENT_CALL_REFUSED`, written by the admission itself, so every door records it the same way). Counted by member, "
|
|
1055
|
+
+ "with the distribution of `door` reported: the refusals by the door they came through. A decision row carries no door, and "
|
|
1056
|
+
+ "neither does a refusal of a call made outside a conversation or a flow run (an application's own call, a deliberation "
|
|
1057
|
+
+ "outside a flow), so the no-door tally is those refusals plus the decision rows. `reason` is not the field reported: on a "
|
|
1058
|
+
+ "decision row it is the administrator's own words.",
|
|
1059
|
+
emittedCountLimit: {
|
|
1060
|
+
unreached: "a count shows the rows that were written, not that each suspension, resumption and refused call left one, since both rows are "
|
|
1061
|
+
+ "written best-effort; nor that every call to a suspended agent was refused, which is C5's question",
|
|
1062
|
+
absenceUnreadable: "a decision whose post-commit write failed, or a refused call whose fire-and-forget write failed, leaves no row, exactly as a "
|
|
1063
|
+
+ "suspension nobody made or a suspended agent nobody called; the failure is logged, not recorded",
|
|
1064
|
+
},
|
|
1065
|
+
requiredAuditEventMembers: [CoreAuditableEventType.AGENT_SUSPENSION_CHANGED, CoreAuditableEventType.SUSPENDED_AGENT_CALL_REFUSED],
|
|
1066
|
+
reportedAuditEventDataDistribution: { field: "door" },
|
|
1067
|
+
},
|
|
1068
|
+
{
|
|
1069
|
+
clauseRef: "C5",
|
|
1070
|
+
obligation: "A suspended agent's call is refused before its model runs at each of the agent service's three entry points, for each "
|
|
1071
|
+
+ "organization the call works for (the one it names, the tenant of the case it works on, the organization of its work frame "
|
|
1072
|
+
+ "and that of the conversation or flow run it is made in) and not another, application-wide for every call, for a "
|
|
1073
|
+
+ "representing organization's suspension too; a call that works for no organization answers to the application's suspensions "
|
|
1074
|
+
+ "only; resuming lets calls run; an unreadable suspension, an unreadable case, or an application whose id cannot be resolved "
|
|
1075
|
+
+ "refuses the call as retryable; an application's boot arms the admission before it serves anything, and a process that boots "
|
|
1076
|
+
+ "no application reads no suspension; an organization's administrator may suspend for their own organization only and never "
|
|
1077
|
+
+ "application-wide; and a fresh or resumed turn ends failed under AGENT_SUSPENDED.",
|
|
1078
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1079
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1080
|
+
homeRationale: "A refused call reaches no model, so the refusal is evidenced by driving it, with the real admission and only the model stubbed. "
|
|
1081
|
+
+ "Driven by tests: the admission at its three entry points; the in-app, A2A, autonomous and chat-platform doors, each through "
|
|
1082
|
+
+ "the shared conversational runtime under its own initiation kind, ending failed with a refusal row naming its door; a flow's "
|
|
1083
|
+
+ "`agent.call` step and a step operation's own call, through the real flow service (door `flow_step`); a deliberation inside "
|
|
1084
|
+
+ "an organization's work; an application's own call inside an actor run on a case; the application's boot arming the "
|
|
1085
|
+
+ "admission; the verbs' authority and the admission's reader on a real MongoDB and PostgreSQL; and the audit rows' severity "
|
|
1086
|
+
+ "and routing. Resting on the shared entry points, not driven themselves: the A2A controller, the chat-command door and the "
|
|
1087
|
+
+ "autonomous-turn service, so no live A2A client over HTTP, no live Slack or Teams delivery and no scheduled autonomous "
|
|
1088
|
+
+ "occurrence. Each driven test carries an `assurance-evidence:` marker naming this clause.",
|
|
1089
|
+
},
|
|
1090
|
+
],
|
|
1091
|
+
minimumEvidence: "C1: per agent on an AI capability of the contracted service whose declared use is not NO_DECISION, its consequence, its stated "
|
|
1092
|
+
+ "review, its changing tools split into reviewer-routed, owner-routed and ungated, and the capability-level uses with a consequence. "
|
|
1093
|
+
+ "C2: the decision rows of the period, by member. C4: the suspension decisions and the refused calls of the period, by member, "
|
|
1094
|
+
+ "with the refused calls by door and the application-wide decisions' justifications.",
|
|
1095
|
+
verificationSteps: [
|
|
1096
|
+
"Read the agents and their declared uses, and derive the contracted-service AI capabilities with the engine's partition.",
|
|
1097
|
+
"For each use declared to affect a person, derive its oversight from its tools' approval gates.",
|
|
1098
|
+
"Name every such use whose agent changes data with no gate, and every capability-level use whose review is stated only.",
|
|
1099
|
+
"Count the decision rows of the period by member.",
|
|
1100
|
+
"Read the suspension decisions and the refused calls of the period with their doors, and confirm each application-wide decision carries its justification.",
|
|
1101
|
+
"Run the marked driven tests for C5, and read which doors they drive and which rest on the shared entry points.",
|
|
1102
|
+
],
|
|
1103
|
+
severityGuidance: "A use declared to affect a person with no stated review is a MAJOR (the declaration is invalid). An agent whose use affects a "
|
|
1104
|
+
+ "person and that changes data with no approval gate is an OBSERVATION: its stated review is the operator's word, and a reviewer "
|
|
1105
|
+
+ "must read whether it covers those tools. A capability-level use reviewed only in words is an OBSERVATION. No decision row in a "
|
|
1106
|
+
+ "period when agents ran gated tools is an OBSERVATION: either nothing was gated, or the evidence export is missing. A "
|
|
1107
|
+
+ "suspended agent whose call reaches a model, or an organization's administrator able to suspend for another organization or "
|
|
1108
|
+
+ "application-wide, is a MAJOR found by C5. An application-wide suspension without its justification is a MAJOR.",
|
|
1109
|
+
limits: "It evidences the gates and the decisions recorded, not that a reviewer read what they approved. It reads tool approval gates only: "
|
|
1110
|
+
+ "a write reviewed through an actor's own question route (`actorExecutionContinuations.decisionRoutes`) reads as ungated here, and "
|
|
1111
|
+
+ "is for the reviewer to weigh. A stop leaves no record (C3). A suspension does not stop a call already running: it ends after "
|
|
1112
|
+
+ "its bounded steps, and the next call is refused. A queued autonomous run is refused at its first model call, not skipped. "
|
|
1113
|
+
+ "The dev companion and the platform's services, which boot no application, answer to no suspension, by design. A call that "
|
|
1114
|
+
+ "works for no organization (an application's own call outside any request, turn, flow or case) answers to the application's "
|
|
1115
|
+
+ "suspensions only. The person sees AGENT_SUSPENDED on a live turn only; a reloaded conversation or a chat platform shows the "
|
|
1116
|
+
+ "turn as interrupted until the code is stored (#2827). An A2A client receives `-32018` only on a blocking `message/send` or "
|
|
1117
|
+
+ "`message/stream` answer; one that reads a stored failed task later sees a plain failure (#3020). The A2A, chat-platform and "
|
|
1118
|
+
+ "autonomous doors are driven through the shared conversational runtime rather than their own controllers, and none is driven "
|
|
1119
|
+
+ "live. Both C4 rows are best-effort: a refused call's record is fire-and-forget and a decision's is written after the change "
|
|
1120
|
+
+ "commits, so a failed write leaves no row and is only logged.",
|
|
1121
|
+
},
|
|
1122
|
+
"evidence-requirement:ai-produced-change-declaration": {
|
|
1123
|
+
schemaVersion: 1,
|
|
1124
|
+
applicability: "Applies to every application. Its population is every coding-agent provider that can write the application's code (an authored "
|
|
1125
|
+
+ "execution profile, or the build-time companion scope admitting the agent protocol), every commit in the application's history, "
|
|
1126
|
+
+ "and the decided elements a coding brief carries that a research step filled. "
|
|
1127
|
+
+ "An application whose code no coding agent wrote has an empty C1 population and no system acceptance in C3, which is the answer "
|
|
1128
|
+
+ "rather than a gap.",
|
|
1129
|
+
clauses: [
|
|
1130
|
+
{
|
|
1131
|
+
clauseRef: "C1",
|
|
1132
|
+
obligation: "Every coding agent that can write the application's code is served by a vendor with an established compliance profile, runs only "
|
|
1133
|
+
+ "in the environments its execution profile admits and never in production, and either asks before writing, so its ceiling and "
|
|
1134
|
+
+ "its task's write scope hold for every declared edit, or is declared unrestricted, because nothing can hold it in flight.",
|
|
1135
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1136
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1137
|
+
homeRationale: "The vendors, the execution profiles and their ceilings and environments are the application's committed configuration, and each "
|
|
1138
|
+
+ "agent's escalation behaviour is its provider's measured declaration on the engine contract. Whether the guards can fire is the "
|
|
1139
|
+
+ "engine's own predicate over those two, so the commit is the evidence of what an agent may do. Whether the host sets a "
|
|
1140
|
+
+ "configuration pointer is a host fact the dispatch checks at every run, and is not read here.",
|
|
1141
|
+
},
|
|
1142
|
+
{
|
|
1143
|
+
clauseRef: "C2",
|
|
1144
|
+
obligation: "A permission request or client write outside the ceiling or the task's write scope is refused, a ceiling narrower than unrestricted "
|
|
1145
|
+
+ "is refused before a session opens for an agent its guards cannot hold, a run is refused outside the environments its profile "
|
|
1146
|
+
+ "admits, a coding provider with no established compliance profile is refused at composition, and an automatic acceptance is "
|
|
1147
|
+
+ "committed under the companion's identity.",
|
|
1148
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1149
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1150
|
+
homeRationale: "Each refusal leaves no row: the runner answers the agent, the dispatch throws before spawning, and composition throws at boot. "
|
|
1151
|
+
+ "They are evidenced by driving them, and the driven tests carry `assurance-evidence:` markers naming this clause.",
|
|
1152
|
+
},
|
|
1153
|
+
{
|
|
1154
|
+
clauseRef: "C3",
|
|
1155
|
+
obligation: "Every change accepted from the creation lane is committed under the identity of whoever accepted it: the companion's for an "
|
|
1156
|
+
+ "acceptance the system made under the autonomy posture, a person's for theirs. Each names its origin, an automatic one the "
|
|
1157
|
+
+ "posture it ran under, and no system acceptance is made under a posture that keeps acceptance with a person.",
|
|
1158
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1159
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1160
|
+
homeRationale: "Acceptance IS the commit, so the application's committed history is the record, and it is DECLARED in the sense the home means: "
|
|
1161
|
+
+ "the committed artifact is the evidence and one query reads it. The plane is PRODUCTION because the history covers a period and "
|
|
1162
|
+
+ "shows the attribution operated, which no configuration can. The collector reads it at HEAD, scoped to the application's path, "
|
|
1163
|
+
+ "and every commit git counts there is parsed.",
|
|
1164
|
+
},
|
|
1165
|
+
{
|
|
1166
|
+
clauseRef: "C4",
|
|
1167
|
+
obligation: "What each run was refused, and what it changed outside its task's scope, is kept for the period the claim covers.",
|
|
1168
|
+
home: AssuranceEvidenceHome.UNPRODUCED,
|
|
1169
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1170
|
+
homeRationale: "A run's permission decisions and its out-of-scope paths are recorded on the run's result, which is the companion's per-run state: "
|
|
1171
|
+
+ "deleting it costs a re-run and no fact about the product, so nothing retains it. UNPRODUCED is the honest value until a durable "
|
|
1172
|
+
+ "record of a run's refusals exists.",
|
|
1173
|
+
},
|
|
1174
|
+
{
|
|
1175
|
+
clauseRef: "C5",
|
|
1176
|
+
obligation: "Every section of a coding agent's brief declares who wrote it, TRUSTED or THIRD_PARTY, and text written outside the repository "
|
|
1177
|
+
+ "travels only in THIRD_PARTY sections, never inlined in a trusted one, for the population the compiled catalog derives: the "
|
|
1178
|
+
+ "outputs of research steps and of the other children of every composite that admits one, at the grain each output declares. "
|
|
1179
|
+
+ "Text a later step wrote after reading that population is not in it, and travels as trusted (#2937). When the catalog cannot "
|
|
1180
|
+
+ "say which elements came from the web, the served elements and the elements the website plan resolves travel as THIRD_PARTY, "
|
|
1181
|
+
+ "marked and never blocked; the vision one-liner, the task labels, the domain-plan slice, the placement plan and the build-task "
|
|
1182
|
+
+ "objective are not routed. The application declares whether a content-safety classifier serves its build-time companion scope.",
|
|
1183
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1184
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1185
|
+
homeRationale: "The trust class is a required field of every brief section the companion assembles, so a section without one does not compile. "
|
|
1186
|
+
+ "Which elements are third-party is derived from the compiled playbook catalog the application's companion ships (a composite "
|
|
1187
|
+
+ "admitting a research child taints its other children's outputs, at the grain each output declares). The classifier binding "
|
|
1188
|
+
+ "is the application's committed companion provider scope. All three are committed or packaged artifacts, read by one query.",
|
|
1189
|
+
},
|
|
1190
|
+
{
|
|
1191
|
+
clauseRef: "C6",
|
|
1192
|
+
obligation: "A third-party element a bound classifier flags is withheld from the brief with a notice; withholding is scoped to the brief, so "
|
|
1193
|
+
+ "the element stays in the specifications and the `wildo context elements` door still returns it unscreened (#2938). One no "
|
|
1194
|
+
+ "classifier screened (none bound, or the bound one could not answer) reaches the brief marked as unscreened, never unmarked. Up "
|
|
1195
|
+
+ "to four screening calls are in flight at once; once one comes back UNSCREENED_CLASSIFIER_UNAVAILABLE, for any failure to screen "
|
|
1196
|
+
+ "other than the element's own length (the seam does not distinguish a transport failure from a refusal, an unavailable vendor or "
|
|
1197
|
+
+ "a misconfigured contract), the brief's elements not yet sent carry that outcome without a call. "
|
|
1198
|
+
+ "An element too long for the seam marks only itself. A catalog that cannot say which elements came from the web routes the "
|
|
1199
|
+
+ "served and website-plan elements to third-party sections. A third-party source with no screening outcome cannot be composed "
|
|
1200
|
+
+ "into a coding task, and the coding contract cannot represent one.",
|
|
1201
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1202
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1203
|
+
homeRationale: "Each outcome is a driven result: the companion's brief-trust suite drives a planted injection through the real content-safety "
|
|
1204
|
+
+ "service over a stubbed classifier endpoint, and the runtime and contract suites drive the refusals. The planted injection "
|
|
1205
|
+
+ "starts from a COMMITTED element; the path from a fetched page through the evidence to the producing step is not driven, and is "
|
|
1206
|
+
+ "#2938's. The driven tests carry `assurance-evidence:` markers naming this clause.",
|
|
1207
|
+
},
|
|
1208
|
+
{
|
|
1209
|
+
clauseRef: "C7",
|
|
1210
|
+
obligation: "What third-party text each run's brief carried, and what screening concluded about it, is kept for the period the claim covers.",
|
|
1211
|
+
home: AssuranceEvidenceHome.UNPRODUCED,
|
|
1212
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1213
|
+
homeRationale: "Each run's screening outcomes are recorded on its execution row (`contextSelection.thirdPartyScreenings`: source, outcome, "
|
|
1214
|
+
+ "provider and reason, never the text) and on the contract its checkpoint holds. That is the companion's per-run state, kept for "
|
|
1215
|
+
+ "as long as the companion keeps its execution rows, which is no stated period: deleting it costs a re-run and no fact about "
|
|
1216
|
+
+ "the product. UNPRODUCED, as C4 is, until a durable record of a run's context exists.",
|
|
1217
|
+
},
|
|
1218
|
+
],
|
|
1219
|
+
minimumEvidence: "C1: per coding-agent provider, whether the companion can dispatch it, its compliance profile and deployment model, its measured "
|
|
1220
|
+
+ "escalation, and per execution profile its ceiling, environments, whether its guards can fire (with the configuration pointer it "
|
|
1221
|
+
+ "needs) and its declared model destinations. C3: the application's commits split into system acceptances (by origin and posture), "
|
|
1222
|
+
+ "a person's acceptances through the lane, and a person's own changes, with any commit whose author and origin disagree. C5: the "
|
|
1223
|
+
+ "research steps and composites the catalog declares (none is INSUFFICIENT EVIDENCE: the catalog was misread), the research-informed "
|
|
1224
|
+
+ "families and aspects with the research steps that feed each, any output of a research composite left outside that population, "
|
|
1225
|
+
+ "and whether the companion scope binds a classifier for AI_CONTENT_SAFETY, which one, and whether the capability is enabled.",
|
|
1226
|
+
verificationSteps: [
|
|
1227
|
+
"Read the application's execution profiles and its build-time companion scope, and the engine's agent contracts and compliance profiles.",
|
|
1228
|
+
"For each profile, derive whether the guards can fire with the engine's own predicate, and name every vendor without an established profile and every profile admitting production.",
|
|
1229
|
+
"Read the application's history at HEAD for its path, check every counted commit was parsed, and split it by author and lane trailer.",
|
|
1230
|
+
"Name every companion-authored commit without a lane trailer, every author contradicted by its recorded origin, and every system acceptance recorded under GUIDED.",
|
|
1231
|
+
"Read the compiled playbook catalog the application's companion ships, derive its research-informed families and aspects, and name any output of a research composite left outside them.",
|
|
1232
|
+
"Read the application's companion provider scope and whether AI_CONTENT_SAFETY is enabled, and name the classifier that screens a coding brief's third-party text, or that none does.",
|
|
1233
|
+
],
|
|
1234
|
+
severityGuidance: "C1: a coding vendor with no established compliance profile, or an execution profile admitting production, is a MAJOR. An agent "
|
|
1235
|
+
+ "dispatchable with no in-flight gate, a ceiling the dispatch will refuse, an authored profile the companion scope does not admit, "
|
|
1236
|
+
+ "and an environment beyond development are OBSERVATIONS: each is a stated posture the claim does not cover. C3: a commit under the "
|
|
1237
|
+
+ "companion's identity the accept door did not write, a commit whose author and recorded origin disagree, and a system acceptance "
|
|
1238
|
+
+ "under GUIDED are each a MAJOR, since the history then misstates who accepted the change. System acceptances written before the "
|
|
1239
|
+
+ "posture trailer existed are an OBSERVATION. A system acceptance that records an automatic origin and no posture (the accept "
|
|
1240
|
+
+ "door writes both together), or a posture outside the vocabulary, leaves its posture unknown: INSUFFICIENT EVIDENCE. Under "
|
|
1241
|
+
+ "YOLO or YOLO_LIGHT any share of system acceptances is the posture working. C5: an output of a research composite outside the "
|
|
1242
|
+
+ "research-informed population is a MAJOR, since its text would travel as trusted; a catalog naming no research step or composite "
|
|
1243
|
+
+ "is INSUFFICIENT EVIDENCE; no classifier bound in the companion scope is an OBSERVATION (every third-party element travels marked "
|
|
1244
|
+
+ "UNSCREENED_NO_CLASSIFIER, the stated posture), and so is a selection naming a provider the scope does not serve it with (every "
|
|
1245
|
+
+ "element travels marked UNSCREENED_CLASSIFIER_UNAVAILABLE). C6: a third-party element reaching a brief unmarked is a MAJOR.",
|
|
1246
|
+
limits: "A system acceptance is not a review: under the decided posture the creator is notified and can revert, and a person reviewed only "
|
|
1247
|
+
+ "the changes C3 lists as a person's. A shell command declares no location, so what it writes is recorded after the run and never "
|
|
1248
|
+
+ "refused, under any ceiling, and for an application inside a larger repository a NEW file a shell command creates outside "
|
|
1249
|
+
+ "the application is not listed at all (untracked files are listed from the application root). An agent that never asks "
|
|
1250
|
+
+ "before writing is bound by neither guard, and only the change set records it. "
|
|
1251
|
+
+ "A commit an agent makes itself through a shell command carries the local identity and reads as a person's; the run records that "
|
|
1252
|
+
+ "the history moved (`historyMutated`), and that record is per-run state (C4). Whether the host supplies OpenCode's configuration "
|
|
1253
|
+
+ "pointer is checked at dispatch, not here. C3 counts commits that touch the application's path, so in a repository that holds "
|
|
1254
|
+
+ "more than the application a person's commit there may also carry other work. "
|
|
1255
|
+
+ "Screening is detection, not a guarantee, and it happens where the brief is assembled, AFTER a generation model has already read "
|
|
1256
|
+
+ "the research evidence and written the element. Withholding is scoped to the brief: the flagged element stays in the "
|
|
1257
|
+
+ "specifications, the coding agent's `wildo context elements get` returns it unscreened, and a research step's evidence reaching a "
|
|
1258
|
+
+ "sibling generation step's brief is not screened (#2938). Provenance is recorded per family and aspect (`researchedBy` on every "
|
|
1259
|
+
+ "decided element the door lists, and `contentTrust` stamped on the research turn's evidence files); which sentence came from the "
|
|
1260
|
+
+ "web is recorded nowhere. Taint is not followed through a reader: text a step wrote after reading a research-informed family "
|
|
1261
|
+
+ "travels as trusted (#2937). A research composite whose producing child declares no output aspects taints its whole family, "
|
|
1262
|
+
+ "coarser than needed and never under-marked. C6's planted injection starts from a committed element; the path from a fetched page "
|
|
1263
|
+
+ "to the producing step is #2938's.",
|
|
1264
|
+
},
|
|
1265
|
+
"evidence-requirement:model-spend-limit-record": {
|
|
1266
|
+
schemaVersion: 1,
|
|
1267
|
+
applicability: "Applies to every application whose contracted service reaches an AI capability. An application with none spends no model tokens "
|
|
1268
|
+
+ "through an agent, which is the answer rather than a gap. With billing on, the population is the application's organization plans; "
|
|
1269
|
+
+ "with billing off, it is the default features profile every organization holds.",
|
|
1270
|
+
clauses: [
|
|
1271
|
+
{
|
|
1272
|
+
clauseRef: "C1",
|
|
1273
|
+
obligation: "Every organization plan (or, with billing off, the default features profile) grants a finite organization model allowance per "
|
|
1274
|
+
+ "period (`a2a.llm.tokensPerPeriod`) and a per-person ceiling for in-app chat (`a2a.llm.tokensPerPersonPerPeriod`), and, for each "
|
|
1275
|
+
+ "non-generation unit its organizations' calls spend, a finite per-organization ceiling: embedding tokens "
|
|
1276
|
+
+ "(`ai.embeddings.tokensPerPeriod`), document-extraction pages (`ai.documentExtraction.pagesPerPeriod`) and content-screening text "
|
|
1277
|
+
+ "records (`ai.contentSafety.textRecordsPerPeriod`).",
|
|
1278
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1279
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1280
|
+
homeRationale: "Every limit is a committed declaration (a plan's granted limits, or the default profile's limits), resolved live by feature "
|
|
1281
|
+
+ "resolution and enforced by the engine before every turn of every conversational door, before each model call of a declared "
|
|
1282
|
+
+ "flow, and before every embedding, extraction and screening call, fail-closed. What the application bounds is exactly what it "
|
|
1283
|
+
+ "commits, and the commit is the evidence.",
|
|
1284
|
+
},
|
|
1285
|
+
{
|
|
1286
|
+
clauseRef: "C2",
|
|
1287
|
+
obligation: "Each turn, and each model call of a declared flow, refused because an organization allowance or a person ceiling is spent, or "
|
|
1288
|
+
+ "could not be verified, is recorded; so is each embedding, extraction and screening call an organization's unit ceiling refused.",
|
|
1289
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1290
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1291
|
+
homeRationale: "A durable row class in the application's own audit store, written by the budget gate itself for every door (A2A, autonomous, "
|
|
1292
|
+
+ "chat platform, in-app, and a declared flow's step), so the doors cannot disagree. Each row names its door, the limit that refused and the reason; the "
|
|
1293
|
+
+ "autonomous door also writes its own refusal row, which carries more than spend.",
|
|
1294
|
+
emittedCountLimit: { unreached: "a count shows refusals were recorded, not that every spent limit refused: whether a limit was granted is C1's question" },
|
|
1295
|
+
requiredAuditEventMembers: [CoreAuditableEventType.MODEL_SPEND_LIMIT_REFUSED],
|
|
1296
|
+
// By door: the five doors share one gate, so a door with no refusal while its siblings refuse is the reading this
|
|
1297
|
+
// tally makes possible (`flow_step` is a door's value since #2935; earlier rows carry the four conversational ones). The limit that refused (`ceiling`) is on every row too and read from the rows themselves.
|
|
1298
|
+
// A unit-ceiling refusal of a call made in no door (#3030: the embedding drain, an ingestion's extraction) carries no
|
|
1299
|
+
// `initiationKind`, so it is tallied as absent there and named by its `ceiling`.
|
|
1300
|
+
reportedAuditEventDataDistribution: { field: "initiationKind" },
|
|
1301
|
+
},
|
|
1302
|
+
{
|
|
1303
|
+
clauseRef: "C3",
|
|
1304
|
+
obligation: "An in-app turn past a spent organization allowance, or past a spent person ceiling while the organization still has allowance, is "
|
|
1305
|
+
+ "refused before it runs, a colleague is not, and an unreadable counter refuses rather than admits. The other three doors call the "
|
|
1306
|
+
+ "same gate, so this proves the gate; their call sites are evidenced by their own door tests, which are not marked here.",
|
|
1307
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1308
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1309
|
+
homeRationale: "A refusal before the turn runs leaves no turn behind to read, and whether a colleague is spared is a comparison between two "
|
|
1310
|
+
+ "principals: both are evidenced by driving them against a real store, and the driven tests carry `assurance-evidence:` markers "
|
|
1311
|
+
+ "naming this clause.",
|
|
1312
|
+
},
|
|
1313
|
+
{
|
|
1314
|
+
clauseRef: "C4",
|
|
1315
|
+
obligation: "A declared flow's model call for an organization whose allowance is spent is refused before the model runs, the run ends FAILED "
|
|
1316
|
+
+ "on that step with the refusal on it, and one refusal row names the door `flow_step`; a call a step's resource operation makes "
|
|
1317
|
+
+ "is refused the same way; and a step whose organization is under its allowance is charged to the organization's counter, not "
|
|
1318
|
+
+ "to the requester's ceiling.",
|
|
1319
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1320
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1321
|
+
homeRationale: "A refused call leaves no provider call and no step output behind to read, so it is evidenced by driving it. The flow-service "
|
|
1322
|
+
+ "tests run the real flow service, the real agent service's spend chokepoint and the real budget gate; stubbed are the model "
|
|
1323
|
+
+ "call (`executeAgent`) and the agent's resolution and registry, the actor execution (`runActorExecution`, so no work frame is "
|
|
1324
|
+
+ "opened), the requester's rebuilt authority and execution context (`rebuildDecisionRequesterAuthority`, "
|
|
1325
|
+
+ "`buildContextFromSocketAuth`), the authorizer, the operation dispatch (`executeCustomOperation`), the system reads, the email "
|
|
1326
|
+
+ "service, and the gate's counter store, feature resolution and audit writer. The nested call's real path, a refusal raised "
|
|
1327
|
+
+ "inside an operation's custom implementation and surviving the resource service tier over MongoDB, is pinned by "
|
|
1328
|
+
+ "`model-spend-refusal-through-operation.mongo.integration.test.ts`. Both carry `assurance-evidence:` markers naming this clause.",
|
|
1329
|
+
},
|
|
1330
|
+
{
|
|
1331
|
+
clauseRef: "C5",
|
|
1332
|
+
obligation: "An embedding, a document extraction or a content screening for an organization whose ceiling for that unit is spent, or could "
|
|
1333
|
+
+ "not be verified, is refused before its provider is called, and one refusal row names the unit's ceiling: the embedding drain "
|
|
1334
|
+
+ "leaves that organization's chunks unembedded, not failed, and drains every other organization in the same run; a refused "
|
|
1335
|
+
+ "screening is not-screened, so a guarded agent withholds the text, with the ceiling named rather than a classifier outage; a "
|
|
1336
|
+
+ "refused extraction is reported as a provider that was not reached, never as an unreadable document; what a call consumed is "
|
|
1337
|
+
+ "charged to its organization's counter for that unit; and a ceiling that is not granted refuses nothing.",
|
|
1338
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1339
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1340
|
+
homeRationale: "A refused call leaves no provider call and no vector, text or verdict behind to read, so it is evidenced by driving it. The "
|
|
1341
|
+
+ "budget-service tests run the real gate and accumulator over a stubbed counter store and feature resolution; the embeddings, "
|
|
1342
|
+
+ "extraction and content-safety tests run each real seam with the real gate in front of a stubbed provider; the drain test runs "
|
|
1343
|
+
+ "the real drain over a stubbed chunk store whose embeddings stub answers a spent organization with the ceiling's own error, "
|
|
1344
|
+
+ "built by the function the real seam throws through; the retrieval test runs the real retrieval service the same way; and the "
|
|
1345
|
+
+ "PostgreSQL store's exclusion of a spent organization is run against a real server. Each carries an `assurance-evidence:` "
|
|
1346
|
+
+ "marker naming this clause.",
|
|
1347
|
+
},
|
|
1348
|
+
],
|
|
1349
|
+
minimumEvidence: "C1: whether billing is on; per organization plan (billing on) or for the default profile (billing off), the granted organization "
|
|
1350
|
+
+ "allowance, per-person ceiling and three unit ceilings (embedding tokens, extraction pages, screening text records), unlimited "
|
|
1351
|
+
+ "shown as unlimited and absent as absent; the plans that grant none. C2: the count of spend refusals over the period, by door.",
|
|
1352
|
+
verificationSteps: [
|
|
1353
|
+
"Read whether billing is enabled from the application's configuration.",
|
|
1354
|
+
"With billing on, read the organization-scoped plan products and their granted limits; with billing off, read the default features profile's limits.",
|
|
1355
|
+
"Name every plan (or the profile) granting no organization allowance, granting an unlimited one, and granting no per-person ceiling.",
|
|
1356
|
+
"Name every plan (or the profile) granting no embedding, extraction or screening ceiling while the AI capability that spends it reaches the contracted service and something spends it: a RAG-source field for embeddings, a RAG-source file field sent to a provider for extraction.",
|
|
1357
|
+
"Read the spend-refusal rows of the period from the application audit door, by door.",
|
|
1358
|
+
],
|
|
1359
|
+
severityGuidance: "While an AI capability reaches the contracted service, a plan or the billing-off profile leaving the organization allowance "
|
|
1360
|
+
+ "unbounded (not granted, or granted unlimited, which resolve the same) is a MAJOR whatever its per-person ceiling: that ceiling "
|
|
1361
|
+
+ "counts in-app chat only, so the external, autonomous and chat-platform doors and every declared flow spend without bound on the "
|
|
1362
|
+
+ "organization's behalf. "
|
|
1363
|
+
+ "Billing on with no organization plan, or billing off with no default profile, is the same MAJOR: every organization resolves to the "
|
|
1364
|
+
+ "unlimited default. A finite allowance with no per-person ceiling is a MINOR: one account can exhaust it in in-app chat. A plan "
|
|
1365
|
+
+ "granting no ceiling (or an unlimited one) for a unit something spends is a MINOR per unit: those calls are billed outside the "
|
|
1366
|
+
+ "generation allowance and nothing else bounds them. Embeddings are spent only where a field is a RAG source, extraction pages "
|
|
1367
|
+
+ "only where a RAG-source file field sends its files to a provider (`textExtraction.ocr`), and screening wherever a classifier "
|
|
1368
|
+
+ "reaches the contracted service, since an organization's policy can switch it on for any agent at runtime.",
|
|
1369
|
+
limits: "It evidences what the plans grant, not that an organization holds a plan: an organization with no subscription resolves to the "
|
|
1370
|
+
+ "definition default, which is unlimited. Consumption is counted after each model call, so a limit can be overrun by the turns and "
|
|
1371
|
+
+ "flow calls in flight when it is reached, and by a turn resumed from a person's decision, which is not asked again (the pause may "
|
|
1372
|
+
+ "cross a period). Counted at an ESTIMATE, not a report: a call that succeeded and whose provider reported no usage (four characters "
|
|
1373
|
+
+ "to a token, 6,240 tokens per image; the trace does not carry the compiled system prompt or a tool loop's tools, so a text call is "
|
|
1374
|
+
+ "under-estimated); a report the provider marked incomplete is charged the larger of the two; a coding agent's transport that "
|
|
1375
|
+
+ "reports no token counter is charged its `contextUsed`, a lower bound that under-counts every model turn after the first and all "
|
|
1376
|
+
+ "output. A call that failed and reported no usage is charged nothing. Counted but NOT refused: a model call made outside any turn "
|
|
1377
|
+
+ "or flow (the application's backend code, an inbound handler, a resource operation an MCP tool dispatches), charged to the "
|
|
1378
|
+
+ "organization it is attributed to (an explicit attribution, the work frame's tenant, a work-unit case's row) and to nobody when it "
|
|
1379
|
+
+ "is attributed to none, and asked nothing first (#2957); and a job a flow enqueues, which runs outside the flow's spend frame on "
|
|
1380
|
+
+ "every queue adapter and is counted the same way. NOT counted against the generation allowance: RAG embeddings, document "
|
|
1381
|
+
+ "extraction and content screening, billed per input token, page or text record; each is bounded by its own per-organization "
|
|
1382
|
+
+ "ceiling instead (#3030), asked before every call wherever it is made, and counted after it (an embedding with no reported usage at "
|
|
1383
|
+
+ "four characters to a token; an extraction at the pages the provider processed; a screening at the text records it answered "
|
|
1384
|
+
+ "for), so calls in flight can overrun a unit ceiling too; a call for no organization (the dev companion, platform tooling) is "
|
|
1385
|
+
+ "asked nothing and counted nowhere; a spent embedding ceiling also stops a retrieval's query embedding, which then searches by "
|
|
1386
|
+
+ "wording only. A manual per-organization override is not read. A person's ceiling counts in-app chat only; their Slack or Teams messages "
|
|
1387
|
+
+ "pay the organization allowance. The A2A burst throttle bounds turns per minute, not spend; it is `WILDO.OPERATIONS.AUTHENTICATED_BURST_LIMIT`'s evidence, not this objective's. "
|
|
1388
|
+
+ "MCP tools dispatch resource operations, not turns: an agent such an operation calls is a call outside any turn. The refusal row is written fire-and-forget: a failed "
|
|
1389
|
+
+ "write is logged, not retried, and the engine's own container always binds the gate and the audit service; a host that composes a "
|
|
1390
|
+
+ "container without the gate enforces no limit, and one without the audit service refuses with no row.",
|
|
1391
|
+
},
|
|
1392
|
+
"evidence-requirement:authenticated-burst-limit-record": {
|
|
1393
|
+
schemaVersion: 1,
|
|
1394
|
+
applicability: "Applies to every application: the engine mounts the throttle on both external agent transports, and the member rates on every "
|
|
1395
|
+
+ "linked chat platform. An application that exposes neither transport over the period answers zero rows, which evidences nothing "
|
|
1396
|
+
+ "either way.",
|
|
1397
|
+
clauses: [
|
|
1398
|
+
{
|
|
1399
|
+
clauseRef: "C1",
|
|
1400
|
+
obligation: "Every window in which an A2A or MCP caller's owner passed its bound leaves one `EXTERNAL_TURN_RATE_LIMITED` row naming the "
|
|
1401
|
+
+ "transport, the owner counted, the window and the limit as declared and as enforced, with the refused principal as the row's "
|
|
1402
|
+
+ "actor and its organization as the row's tenant.",
|
|
1403
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1404
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1405
|
+
homeRationale: "A durable row class in the application's own audit store, written by the throttle itself rather than by either transport, so "
|
|
1406
|
+
+ "the two cannot record a refusal differently; read through the application audit door. Only a window's first refusal is a "
|
|
1407
|
+
+ "row: `INCR` is atomic, so exactly one request per window sees the count one past the limit.",
|
|
1408
|
+
emittedCountLimit: {
|
|
1409
|
+
unreached: "a count shows saturated windows were recorded, not that every saturated window was, nor what each row carries; a window records its first refusal only",
|
|
1410
|
+
absenceUnreadable: "a saturated window whose fire-and-forget write failed leaves no row, exactly as a window nobody saturated",
|
|
1411
|
+
},
|
|
1412
|
+
requiredAuditEventMembers: [CoreAuditableEventType.EXTERNAL_TURN_RATE_LIMITED],
|
|
1413
|
+
// By transport: A2A turns and MCP tool calls carry distinct caps (model spend against database load), so one
|
|
1414
|
+
// transport throttling while the other never does is the reading this tally makes possible.
|
|
1415
|
+
reportedAuditEventDataDistribution: { field: "resourceServer" },
|
|
1416
|
+
},
|
|
1417
|
+
{
|
|
1418
|
+
clauseRef: "C2",
|
|
1419
|
+
obligation: "An A2A or MCP caller past its bound is answered 429 with Retry-After before anything is dispatched; a principal driven past the "
|
|
1420
|
+
+ "bound leaves exactly one row for the window and a colleague under it none; the row carries the declared and the enforced "
|
|
1421
|
+
+ "limit; a counter-store outage refuses nothing and records nothing; and a record write that fails, rejects or never settles "
|
|
1422
|
+
+ "never delays or fails the refusal. A call past its operation's own declared window, on an MCP tool call, an agent's "
|
|
1423
|
+
+ "function tool or a chat command, is refused before dispatch (429 on MCP, a tool error or a refused command elsewhere); "
|
|
1424
|
+
+ "the window's first refusal, and only it, leaves one `OPERATION_RATE_LIMIT_REFUSED` row naming the plane and the subject counted (a "
|
|
1425
|
+
+ "delegated person as the person); a second credential of the same organization, under the window, leaves none; a write "
|
|
1426
|
+
+ "that rejects, never settles or throws, or an absent audit service, still refuses; and a counter outage fails the call and "
|
|
1427
|
+
+ "records nothing. On an operation's own HTTP route each request is counted once: a signed-in person or machine past the "
|
|
1428
|
+
+ "declared window is refused once authorized and counted per caller, so one person on one address leaves exactly one "
|
|
1429
|
+
+ "`http_route` row and colleagues behind one address each keep their own budget; a visitor, a public request and a link "
|
|
1430
|
+
+ "token, whatever userId its URL names, are counted per address and leave no row.",
|
|
1431
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1432
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1433
|
+
homeRationale: "A refused request leaves no dispatch behind to read, and one row per window against none for a colleague is a comparison "
|
|
1434
|
+
+ "between principals, so the gate is evidenced by driving it: the throttle against a fixed-window counter, the same throttle "
|
|
1435
|
+
+ "over a real Redis writing into a real audit store (A2A through the service's own gate, MCP through the throttle with its "
|
|
1436
|
+
+ "caps), and each transport's controller answering 429 without dispatching. The declared limit is driven the same way: the "
|
|
1437
|
+
+ "limiter's report of each window's first refusal, the MCP server's, the function-tool seam's and the chat command door's "
|
|
1438
|
+
+ "recording of it, the agent-tool planes handing their tools the audit service, and a real Redis and the real limiter "
|
|
1439
|
+
+ "writing into a real audit store through `callTool` and through the operation-dispatch seam (#2994, #3008). On the HTTP "
|
|
1440
|
+
+ "route, the controller's two counts run in their real order with the real limiter, and the same two helpers run over a "
|
|
1441
|
+
+ "real Redis into a real audit store (#2949). Each carries an `assurance-evidence:` marker naming this clause.",
|
|
1442
|
+
},
|
|
1443
|
+
{
|
|
1444
|
+
clauseRef: "C3",
|
|
1445
|
+
obligation: "A linked member past their chat message, command or search rate is refused with a reply before the turn or the command runs, "
|
|
1446
|
+
+ "each counted in the member's own bucket at the documented rate; a refused command or search leaves its run row; and an "
|
|
1447
|
+
+ "unreadable message or command counter refuses nothing and has the delivery tried again, while an unreadable search counter "
|
|
1448
|
+
+ "fails the search with no row, so the person is asked to try again.",
|
|
1449
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1450
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1451
|
+
homeRationale: "A refused chat message runs no turn and leaves no row, so it can only be evidenced by driving it; the chat doors' tests drive "
|
|
1452
|
+
+ "each rate against the door, with the counter, the store and the audit writer stubbed, and carry `assurance-evidence:` "
|
|
1453
|
+
+ "markers naming this clause.",
|
|
1454
|
+
},
|
|
1455
|
+
{
|
|
1456
|
+
clauseRef: "C4",
|
|
1457
|
+
obligation: "Each door declares a finite, positive cap and window: A2A turns and MCP tool calls per owner, and a linked member's chat "
|
|
1458
|
+
+ "messages, commands and searches per member; and the relaxation that raises them applies in development and test only, so "
|
|
1459
|
+
+ "staging and production enforce the declared number.",
|
|
1460
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1461
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1462
|
+
homeRationale: "The caps and windows are committed constants of the engine's burst-limit contract, and the relaxation is a total function "
|
|
1463
|
+
+ "of the declared runtime environment, so reading the contract and running the rule over every environment the engine "
|
|
1464
|
+
+ "names answers the clause; no occurrence evidences a bound. Both are read through "
|
|
1465
|
+
+ "`@wildo-ai/saas-backend-lib/assurance-derivations` from the engine snapshot the application resolves, so the numbers are "
|
|
1466
|
+
+ "the ones its doors enforce. They are engine-authored: no application sets them. An operation's own declared limit is not "
|
|
1467
|
+
+ "read here: the application authors it per operation, and C5 evidences its refusals.",
|
|
1468
|
+
},
|
|
1469
|
+
{
|
|
1470
|
+
clauseRef: "C5",
|
|
1471
|
+
obligation: "Every declared window of an operation's own rate limit that a client passed, on an MCP tool call, an agent's function tool, "
|
|
1472
|
+
+ "a chat command or, for a signed-in person or machine, the operation's own HTTP route, leaves one "
|
|
1473
|
+
+ "`OPERATION_RATE_LIMIT_REFUSED` row naming the plane, the operation, the subject counted (the machine credential, else the "
|
|
1474
|
+
+ "person: through a client they delegated to, in the application, as a linked chat member, or signed in), the window and "
|
|
1475
|
+
+ "the limit as declared and as enforced, with the refused principal as the row's actor and its organization as the row's "
|
|
1476
|
+
+ "tenant.",
|
|
1477
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1478
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1479
|
+
homeRationale: "A durable row class in the application's own audit store, written from the limiter's own report of a window's first "
|
|
1480
|
+
+ "refusal (`INCR` is atomic, so exactly one call per window instance sees the count one past the limit) and read through "
|
|
1481
|
+
+ "the application audit door. A different event from C1's on purpose: this limiter is the application's statement about "
|
|
1482
|
+
+ "one operation, counted per client and failing closed, and sharing C1's event would make its count mean either. "
|
|
1483
|
+
+ "One plane is not one surface: `agent_tool` holds an A2A turn's, an application WebSocket turn's and a chat "
|
|
1484
|
+
+ "conversation turn's tool calls, and the engine's own work-unit case reads for an agent turn, in one bucket.",
|
|
1485
|
+
emittedCountLimit: {
|
|
1486
|
+
unreached: "a count shows saturated declared windows were recorded, not that every saturated window was, nor what each row carries; a window records its first refusal only",
|
|
1487
|
+
absenceUnreadable: "a saturated declared window whose fire-and-forget write failed leaves no row, exactly as a window nobody saturated, and so does a window first refused in a call that met a counter outage",
|
|
1488
|
+
},
|
|
1489
|
+
requiredAuditEventMembers: [CoreAuditableEventType.OPERATION_RATE_LIMIT_REFUSED],
|
|
1490
|
+
// By plane: an MCP client's, an agent's, a chat member's and a signed-in caller's route refusals are four populations, and
|
|
1491
|
+
// one plane refusing while another never does is the reading this tally makes possible. The subject kind is on every row for a reviewer.
|
|
1492
|
+
reportedAuditEventDataDistribution: { field: "plane" },
|
|
1493
|
+
},
|
|
1494
|
+
],
|
|
1495
|
+
minimumEvidence: "C1: the count of `EXTERNAL_TURN_RATE_LIMITED` rows over the period, by transport. C5: the count of `OPERATION_RATE_LIMIT_REFUSED` "
|
|
1496
|
+
+ "rows over the period, by plane. C4: the five caps and windows, and the "
|
|
1497
|
+
+ "multiplier the relaxation applies in each environment, read from the engine contract. C4 collects the declared side; "
|
|
1498
|
+
+ "whether a staging or production row's enforced limit equals its declared limit is still a reviewer's read of the rows, "
|
|
1499
|
+
+ "and is not collected.",
|
|
1500
|
+
verificationSteps: [
|
|
1501
|
+
"Read the five committed caps and windows, and the multiplier the relaxation applies in each environment.",
|
|
1502
|
+
"Read the throttle rows of the period from the application audit door, by transport.",
|
|
1503
|
+
"Read the declared-policy refusal rows of the period, by plane.",
|
|
1504
|
+
"On a staging or production deployment, confirm each row's enforced limit equals its declared limit, for both row classes.",
|
|
1505
|
+
"Run the marked driven tests for C2 and C3.",
|
|
1506
|
+
],
|
|
1507
|
+
severityGuidance: "A staging or production row whose enforced limit differs from its declared limit is a MAJOR: the development and test "
|
|
1508
|
+
+ "relaxation reached an environment that enforces the declared number. "
|
|
1509
|
+
+ "More than one row for one owner, transport and window is an OBSERVATION: noise, not loss. A transport or a chat "
|
|
1510
|
+
+ "door that dispatches before its throttle, and a saturated window with no row while the audit service is bound, are each a "
|
|
1511
|
+
+ "MAJOR, found by C2 and C3 rather than by the rows. A cap or window that is not a finite positive number, a cap that is "
|
|
1512
|
+
+ "not a whole number of calls, and a relaxation in any environment but development and test, are each a MAJOR found by "
|
|
1513
|
+
+ "C4. A cap or window C4 could not read as a number is insufficient evidence, never a MAJOR: it says the contract's "
|
|
1514
|
+
+ "exports moved, not that a door is unbounded. For an operation's declared limit: a saturated declared window with no row "
|
|
1515
|
+
+ "while the audit service is bound is a MAJOR found by C2; a signed-in person or machine counted per address on an HTTP route, "
|
|
1516
|
+
+ "so that a whole address shares one budget and the per-caller count never refuses, is a MAJOR found by C2; a staging or "
|
|
1517
|
+
+ "production row whose enforced limit differs from its declared limit is a MAJOR; more than one row for one plane, subject, "
|
|
1518
|
+
+ "operation and window is an OBSERVATION.",
|
|
1519
|
+
limits: "It evidences a per-window bound and the record of an A2A or MCP window's first refusal, and the record of an operation's own "
|
|
1520
|
+
+ "declared limit on every plane that dispatches it, the operation's own HTTP route for a signed-in person or machine "
|
|
1521
|
+
+ "included (C5). The throttle's count is per owner: an organization's "
|
|
1522
|
+
+ "machine credentials share one window, and a person's delegated clients share theirs. A2A task reads and cancellation are not "
|
|
1523
|
+
+ "counted, only turns; on MCP only tool calls are. A counter-store outage leaves the A2A and MCP bound unenforced, with only a "
|
|
1524
|
+
+ "warning log line, and a principal carrying no owner id, which authentication refuses before the throttle, would be admitted "
|
|
1525
|
+
+ "unthrottled. The record is written fire-and-forget: a failed write is logged, not retried, and a container composed without the "
|
|
1526
|
+
+ "audit service refuses with no row. A chat message refused for its rate leaves no audit row, and no chat refusal is an "
|
|
1527
|
+
+ "`EXTERNAL_TURN_RATE_LIMITED` row. An operation's own declared rate limit is evidenced per client, per plane and per declared "
|
|
1528
|
+
+ "window on MCP tool calls, agent function tools, chat commands and a signed-in caller's requests to the operation's own HTTP "
|
|
1529
|
+
+ "route (C5). Its record is written fire-and-forget: a failed write "
|
|
1530
|
+
+ "is logged, not retried, and a plane assembled without the audit service refuses with no row. A hand-written tool that calls "
|
|
1531
|
+
+ "the limiter itself, rather than dispatching through the engine's seam, is charged and leaves no row. An MCP caller carrying "
|
|
1532
|
+
+ "no keyable identity, which the authenticated producers never yield, would be admitted unthrottled with a warning. The "
|
|
1533
|
+
+ "declared limit fails closed on a counter outage, so the call fails rather than passing and nothing is recorded; with several "
|
|
1534
|
+
+ "declared windows, a window already counted in the call that met the outage keeps its count, so if that call was its first "
|
|
1535
|
+
+ "refusal, that window leaves no row. On the HTTP route, a caller who is not a signed-in person or machine (a visitor, a "
|
|
1536
|
+
+ "public request, a link token) is counted per source address instead, and that count records nothing; a request whose "
|
|
1537
|
+
+ "credential does not verify is counted by neither, and only the global HTTP limiter bounds it. Not evidenced: sustained spend (`WILDO.AI.MODEL_SPEND_LIMIT`), the in-app "
|
|
1538
|
+
+ "conversation's own turn limit, and the REST surface's own limits.",
|
|
1539
|
+
},
|
|
1540
|
+
"evidence-requirement:agent-behavior-reports": {
|
|
1541
|
+
schemaVersion: 1,
|
|
1542
|
+
applicability: "Applies to every application: the report and its lifecycle exist wherever agents answer people in the in-app conversation. "
|
|
1543
|
+
+ "An application nobody reported in over the window answers zero reports, which evidences nothing either way.",
|
|
1544
|
+
clauses: [
|
|
1545
|
+
{
|
|
1546
|
+
clauseRef: "C1",
|
|
1547
|
+
obligation: "Each report received, taken on and ended is recorded on the audit trail: received with its category and severity, triaged "
|
|
1548
|
+
+ "with who took it on and whether its triage target was met, closed with its outcome, who reached it and whether its "
|
|
1549
|
+
+ "outcome target was met.",
|
|
1550
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1551
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1552
|
+
homeRationale: "Audit rows in the application's own trail, read through the application audit door, which is also the operator's only "
|
|
1553
|
+
+ "view of reports across organizations (no variant of the report resource admits a cross-tenant read). Each row carries "
|
|
1554
|
+
+ "ids, the category, the severity, the outcome and the deadline outcomes, never the reply, the comment or the note. A "
|
|
1555
|
+
+ "critical report's received row is a CRITICAL row (C6).",
|
|
1556
|
+
emittedCountLimit: {
|
|
1557
|
+
unreached: "that every report has its rows: each row is written after the report's transaction commits, and a failed write is "
|
|
1558
|
+
+ "logged, never retried, so a report can exist with no row; C2's count of the store is the reconciliation",
|
|
1559
|
+
},
|
|
1560
|
+
requiredAuditEventMembers: [
|
|
1561
|
+
CoreAuditableEventType.AGENT_BEHAVIOR_REPORT_RECEIVED,
|
|
1562
|
+
CoreAuditableEventType.AGENT_BEHAVIOR_REPORT_TRIAGED,
|
|
1563
|
+
CoreAuditableEventType.AGENT_BEHAVIOR_REPORT_CLOSED,
|
|
1564
|
+
],
|
|
1565
|
+
// Every row of the three carries the report's severity, so the tally says how many of the window's rows concern a
|
|
1566
|
+
// critical report.
|
|
1567
|
+
reportedAuditEventDataDistribution: { field: "severity" },
|
|
1568
|
+
},
|
|
1569
|
+
{
|
|
1570
|
+
clauseRef: "C2",
|
|
1571
|
+
obligation: "No report is open past its triage target, and none past its outcome target; reports triaged or closed late are counted.",
|
|
1572
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1573
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1574
|
+
homeRationale: "An ENGINE-owned store, `agentBehaviorReports`, read through `GET /assurance/agent-behavior-report-evidence`, which "
|
|
1575
|
+
+ "returns counts and no row: received by severity and category, closed by state and outcome, triaged and closed late "
|
|
1576
|
+
+ "in the window, and open now past each target, over every open report whatever its window. It reads every row it "
|
|
1577
|
+
+ "counts and refuses the window rather than count in part when a row does not read or the population exceeds its "
|
|
1578
|
+
+ "ceiling. The targets are stored on each report at receipt, so a later configuration change cannot move a running "
|
|
1579
|
+
+ "clock. An overdue report is measured here, never escalated: nothing reminds anyone or raises it.",
|
|
1580
|
+
},
|
|
1581
|
+
{
|
|
1582
|
+
clauseRef: "C3",
|
|
1583
|
+
obligation: "A person reports only a reply of their own conversation, and only in the organization that conversation was held in (or, "
|
|
1584
|
+
+ "held in their own context, the one they work in); only that organization's administrators triage it and read the "
|
|
1585
|
+
+ "reported reply; the report points at the reply and copies none of it.",
|
|
1586
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1587
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1588
|
+
homeRationale: "A refusal leaves no report to read, and who may act is a comparison between principals, so it is evidenced by driving "
|
|
1589
|
+
+ "both stores through the service funnel and the authorizer, and the in-app action in its view. The driven tests carry "
|
|
1590
|
+
+ "`assurance-evidence:` markers naming this clause.",
|
|
1591
|
+
},
|
|
1592
|
+
{
|
|
1593
|
+
clauseRef: "C4",
|
|
1594
|
+
obligation: "A report moves received, triaged, then resolved or dismissed, each step once, with an outcome of its own ending and a "
|
|
1595
|
+
+ "note; a remedy sent with a dismissal is refused; a guardrail remedy is refused unless it is a guardrail decision of the report's "
|
|
1596
|
+
+ "organization or, naming none, of the reported conversation.",
|
|
1597
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1598
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1599
|
+
homeRationale: "A refused transition writes nothing, so the state machine and the remedy check are evidenced by driving them on both "
|
|
1600
|
+
+ "stores against a planted trail (a missing event, another organization's, a non-guardrail event, an unscoped guardrail "
|
|
1601
|
+
+ "event of another conversation, and one of this conversation). Marked like C3.",
|
|
1602
|
+
},
|
|
1603
|
+
{
|
|
1604
|
+
clauseRef: "C5",
|
|
1605
|
+
obligation: "A reporter's erasure removes their id and comment and keeps the report and who handled it; a report under a legal hold "
|
|
1606
|
+
+ "is left as it was; the reporter's access export carries the reports they filed, their comment, the outcome and the "
|
|
1607
|
+
+ "remedy link, and withholds the outcome note and who triaged and closed them.",
|
|
1608
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1609
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1610
|
+
homeRationale: "An erasure leaves the row it changed, but whether it changed exactly the reporter's fields, and whether an export "
|
|
1611
|
+
+ "withholds exactly the others' data, is evidenced by driving the erasure door with the severer the startup registers "
|
|
1612
|
+
+ "and the export orchestrator with the column it declares. Marked like C3.",
|
|
1613
|
+
},
|
|
1614
|
+
{
|
|
1615
|
+
clauseRef: "C6",
|
|
1616
|
+
obligation: "A critical report's received row is stored CRITICAL and dispatched as a CRITICAL envelope, and a medium report's is not.",
|
|
1617
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1618
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1619
|
+
homeRationale: "The severity of a stored row is visible, but whether a destination that takes only CRITICAL rows receives the report "
|
|
1620
|
+
+ "is a property of the dispatch, evidenced by driving the audit service into a recording destination. Marked like C3.",
|
|
1621
|
+
},
|
|
1622
|
+
{
|
|
1623
|
+
clauseRef: "C7",
|
|
1624
|
+
obligation: "The report deadlines the application declares are ones startup accepts, or, absent, the engine's day and week.",
|
|
1625
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1626
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1627
|
+
homeRationale: "`agentBehaviorReports.deadlines` is a committed declaration in the backend configuration, refused at startup when it "
|
|
1628
|
+
+ "declares what a report cannot do (a lapse, a warning, escalation rules, pause states, a fraction of a lapse, an "
|
|
1629
|
+
+ "undeclared calendar, an outcome target no later than the triage target). The declaration, or its absence, is the "
|
|
1630
|
+
+ "evidence.",
|
|
1631
|
+
},
|
|
1632
|
+
{
|
|
1633
|
+
clauseRef: "C8",
|
|
1634
|
+
obligation: "A repeat report of the same reply by the same person while an earlier one is open is answered with the open report at its "
|
|
1635
|
+
+ "current status (received or triaged), without who took it on, writes nothing and raises no second received row, "
|
|
1636
|
+
+ "including when the two arrive at once; once the report closes, a new one is taken.",
|
|
1637
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1638
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1639
|
+
homeRationale: "A folded repeat leaves nothing behind by design, so no stored row shows it: only driving the same report twice, in "
|
|
1640
|
+
+ "sequence and at once, on each store, does. Marked like C3.",
|
|
1641
|
+
},
|
|
1642
|
+
],
|
|
1643
|
+
minimumEvidence: "C1: the window's received, triaged and closed rows, with their severities. C2: the door's counts over the window and the "
|
|
1644
|
+
+ "open queue at the moment of the request. C3 to C6 and C8: the driven runs. C7: the committed declaration, or its absence, "
|
|
1645
|
+
+ "with the targets it resolves to and the issues startup would raise.",
|
|
1646
|
+
verificationSteps: [
|
|
1647
|
+
"Read the report evidence door for the window; an empty window is not conformance.",
|
|
1648
|
+
"Read the audit door for the received, triaged and closed rows over the same window, and reconcile them with the door's received, triaged and closed counts.",
|
|
1649
|
+
"Name every report open past a target, and every triage or closure made late.",
|
|
1650
|
+
"Read the committed deadline declaration and the issues startup raises for it.",
|
|
1651
|
+
"Run the marked driven tests: the report lane on both stores, the in-app action, the export orchestrator and the audit severity.",
|
|
1652
|
+
],
|
|
1653
|
+
severityGuidance: "A report open past its outcome target is a MAJOR: the door does not break overdue reports down by severity, so none is "
|
|
1654
|
+
+ "graded lower. A report open past its triage target, or triaged or closed after its target in the window, is a MINOR. A "
|
|
1655
|
+
+ "committed deadline declaration startup refuses is a MAJOR (the application cannot start on it). A received report with "
|
|
1656
|
+
+ "no audit row is a MAJOR (the trail is the operator's only cross-tenant view). A repeat of an open report written as a "
|
|
1657
|
+
+ "second report, or raising a second received row, is a MINOR (one person can fill the queue with one reply). A window "
|
|
1658
|
+
+ "with no report received, triaged, closed or open is an OBSERVATION.",
|
|
1659
|
+
limits: "It evidences that reports are collected, recorded, triaged and closed against declared targets. It does not evidence that "
|
|
1660
|
+
+ "an outcome was right or a remedy effective. Nobody is paged for a critical report (#9): it reaches the organization's SIEM "
|
|
1661
|
+
+ "route when one is configured and the audit-trail provider fan-out when enabled, as every CRITICAL row does, and is "
|
|
1662
|
+
+ "otherwise seen in the triage queue. A signed-in person's reports are bounded per window and per person, never per address "
|
|
1663
|
+
+ "(five a minute, thirty an hour, a hundred a day), and a repeat of an open report of the same reply returns that report "
|
|
1664
|
+
+ "at its current status (#2949); the bound's refusals are evidenced by "
|
|
1665
|
+
+ "`WILDO.OPERATIONS.AUTHENTICATED_BURST_LIMIT` (C5), not here. The audit rows are written "
|
|
1666
|
+
+ "after the commit, best-effort. Reports are collected from the in-app conversation only, not from Slack or Teams, and a "
|
|
1667
|
+
+ "person working in no organization is not offered the action. Guardrail CRITICAL events are not promoted into this queue. "
|
|
1668
|
+
+ "The reporter's access export withholds the outcome note and who triaged and closed the report, under Art. 15(4) GDPR; "
|
|
1669
|
+
+ "the wider gap of other severer-reached rows the export does not reach is #2958. A high-risk system's serious-incident "
|
|
1670
|
+
+ "notification (AI Act Art. 73) is outside the supported scope (#2851), though a report holds what one would be built "
|
|
1671
|
+
+ "from. The operator reads reports across organizations only through the application audit door, under its grant.",
|
|
1672
|
+
},
|
|
1673
|
+
"evidence-requirement:agent-operation-record": {
|
|
1674
|
+
schemaVersion: 1,
|
|
1675
|
+
applicability: "Applies to every application: the engine writes the record for every agent invocation its runtime traces, including one that "
|
|
1676
|
+
+ "failed before a provider answered, whatever the agent declares. An application with no agent answers zero calls, which "
|
|
1677
|
+
+ "evidences nothing either way.",
|
|
1678
|
+
clauses: [
|
|
1679
|
+
{
|
|
1680
|
+
clauseRef: "C1",
|
|
1681
|
+
obligation: "Every agent invocation the runtime traces leaves one operation record beside its provider-call record, naming the model ids "
|
|
1682
|
+
+ "the provider reported, each tool call and the phase it ended at, the guardrail verdicts and the end with its failure code; "
|
|
1683
|
+
+ "an agent call whose provider-call record was written but whose operation record was not is counted.",
|
|
1684
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1685
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1686
|
+
homeRationale: "An ENGINE-owned store, `agentOperationRecords`: one row per agent call, written by the provider-call recorder beside the "
|
|
1687
|
+
+ "call's `providerCallRecords` row and read for a window through `GET /assurance/agent-operation-evidence`, which returns "
|
|
1688
|
+
+ "counts and no row (ends by failure code, models reported, tool calls by phase and error code, the guardrail totals and the "
|
|
1689
|
+
+ "same guardrail counts per agent, which must sum to the totals). It "
|
|
1690
|
+
+ "names no audit member, rightly: a per-call row in the audit trail would drown the decisions about people it records, and "
|
|
1691
|
+
+ "the guardrails' own audit rows are the screening and moderation objectives' evidence.\n\n"
|
|
1692
|
+
+ "**Completeness is a count, not a per-call join.** The door counts the window's agent generation calls in the provider-call "
|
|
1693
|
+
+ "records and reports the shortfall of operation records against them. A row the door cannot parse is not counted, so it "
|
|
1694
|
+
+ "shows as a missing record rather than as a clean one. More records than calls means the shortfall cannot be trusted (a "
|
|
1695
|
+
+ "duplicate record, or a provider-call count the door could not read, which it reports as zero). Records naming no run are "
|
|
1696
|
+
+ "counted apart (`operationsWithoutExecutionRef`): the run is named only when the call site or its work frame names one.",
|
|
1697
|
+
},
|
|
1698
|
+
{
|
|
1699
|
+
clauseRef: "C2",
|
|
1700
|
+
obligation: "No record holds what a tool was given or returned, a prompt, a reply, an error message, the id of a row a tool read, or a person's id.",
|
|
1701
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1702
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1703
|
+
homeRationale: "A row cannot show what it does not hold, and the door returns counts only, so absence is evidenced by driving a call whose "
|
|
1704
|
+
+ "tool returns marked content (an injected instruction, a person's email, a row id) through the real AI SDK and the real "
|
|
1705
|
+
+ "recorder, and reading the stored record back: none of the marked strings is in it, nor the withheld error's message, and "
|
|
1706
|
+
+ "its keys are exactly the projection's. A tool error keeps its engine reference and never its message, pinned on the "
|
|
1707
|
+
+ "observation directly. The tests carry `assurance-evidence:` markers naming this clause.",
|
|
1708
|
+
},
|
|
1709
|
+
{
|
|
1710
|
+
clauseRef: "C3",
|
|
1711
|
+
obligation: "The verdicts in the record are the guardrails' own: a tool result the screen flagged reads `attack_detected`, and a reply "
|
|
1712
|
+
+ "output moderation withheld ends the call `failed` with `OUTPUT_WITHHELD` and the categories it reached.",
|
|
1713
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1714
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1715
|
+
homeRationale: "The record copies what each guard decided and decides nothing, so the evidence is the same drive with a classifier double "
|
|
1716
|
+
+ "that flags the tool result and, in a second call, the reply: the stored record carries each verdict as the guard reached "
|
|
1717
|
+
+ "it. Mutation-checked when the mechanism landed (the screen built without its verdict observer, the failure arm read as a "
|
|
1718
|
+
+ "success, each reddens a pin). Marked like C2.",
|
|
1719
|
+
},
|
|
1720
|
+
{
|
|
1721
|
+
clauseRef: "C4",
|
|
1722
|
+
obligation: "Records are bound to the OPERATIONAL_LOGS retention class from creation, the class and clock of the provider-call record "
|
|
1723
|
+
+ "they join, and the application declares that class's period.",
|
|
1724
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1725
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1726
|
+
homeRationale: "Both bindings are committed resource configurations and the period is the application's committed `dataRetention`, so the "
|
|
1727
|
+
+ "declarations are the evidence. Whether the sweep then deletes past the period is not re-measured here: it is "
|
|
1728
|
+
+ "`retention-window-declaration` C3, whose enforcement records cover every resource bound to a class, this one included.",
|
|
1729
|
+
},
|
|
1730
|
+
],
|
|
1731
|
+
minimumEvidence: "C1: the door's counts over the window — records by end and failure code, by reported model, tool calls by phase, the guardrail "
|
|
1732
|
+
+ "totals and the same guardrail counts per agent, and the agent calls with no record. C2, C3: a driven call with a screened tool result and a judged reply, and one whose "
|
|
1733
|
+
+ "reply is withheld, each read back as one record holding the verdicts and none of the content. C4: the record's and the "
|
|
1734
|
+
+ "provider-call record's retention bindings, and whether the application declares the OPERATIONAL_LOGS period.",
|
|
1735
|
+
verificationSteps: [
|
|
1736
|
+
"Read the agent-operation evidence door for the window; an empty window is not conformance.",
|
|
1737
|
+
"Compare the agent calls in the provider-call records with the operation records, and name the shortfall.",
|
|
1738
|
+
"Drive one agent call whose tool returns a marked string and an injected instruction, with a classifier double flagging it, and read the record back.",
|
|
1739
|
+
"Confirm the record names the provider-reported model, the call at `result` with `screening: attack_detected`, the moderation summary and `end: completed`, and none of the marked strings.",
|
|
1740
|
+
"Drive a reply the classifier withholds; confirm `end: failed`, `failureCode: OUTPUT_WITHHELD`, the withholding's categories, and no text.",
|
|
1741
|
+
"Read the two resources' retention bindings and the application's declared periods.",
|
|
1742
|
+
],
|
|
1743
|
+
severityGuidance: "A window where agent calls were recorded and no operation record exists is a MAJOR: the recorder is not writing, and nothing says "
|
|
1744
|
+
+ "what any agent did. Fewer missing records is a MINOR, naming the count: recording is best-effort and a write can fail. A record "
|
|
1745
|
+
+ "that holds tool output, a prompt, a reply or a person's id is a MAJOR: it is a second copy of what the transcript's "
|
|
1746
|
+
+ "minimisation removes. A record bound to no retention class is a MAJOR (it is kept for the application's whole life); bound to "
|
|
1747
|
+
+ "a different class or clock from the provider-call record it joins, a MINOR (one outlives the other). A record whose provider "
|
|
1748
|
+
+ "reported no model id is an OBSERVATION (the provider reported none, or the call failed before any provider response), and so "
|
|
1749
|
+
+ "is an application that declares no OPERATIONAL_LOGS "
|
|
1750
|
+
+ "period: the rows are kept until it does, and storage limitation owns that finding. A per-agent split that does not sum to "
|
|
1751
|
+
+ "the totals (operations, any guardrail count or any withheld code), or names an agent twice, is refused as unreadable "
|
|
1752
|
+
+ "(insufficient evidence): a record attributed to no agent or to two means the door answered something other than one "
|
|
1753
|
+
+ "accumulation, so no grade read from its totals would mean anything.",
|
|
1754
|
+
limits: "It evidences that each agent call's reported model, tools, guardrail verdicts and end are recorded, and that the record carries "
|
|
1755
|
+
+ "no content. It does not evidence WHY an agent did something, what a tool returned, or what a person did with the answer. A call "
|
|
1756
|
+
+ "made outside the engine's agent runtime (a provider SDK called directly, or a coding agent's own calls to its model) leaves no "
|
|
1757
|
+
+ "record at all, and the shortfall count cannot see it, since it leaves no provider-call record either. The shortfall is a "
|
|
1758
|
+
+ "difference of two counts in one window, so a duplicated record could mask a missing one. A record lists at most 100 tool calls "
|
|
1759
|
+
+ "and counts the rest without listing them. A call whose provider-call record itself failed to write, or found no repository "
|
|
1760
|
+
+ "while the registry was initialising, leaves neither row, and no count shows it. A coding-agent dispatch leaves one record of "
|
|
1761
|
+
+ "the dispatch, with no reported model and no tool list. A record names a run only when the call site or its work frame names "
|
|
1762
|
+
+ "one; the rest are counted under `operationsWithoutExecutionRef`. An inbound delivery refused by "
|
|
1763
|
+
+ "screening before an autonomous turn starts leaves no call and no record; its audit row is the screening objective's evidence. "
|
|
1764
|
+
+ "A record names the run a call was made in, not the person: the path from a person runs through the run, which is kept under "
|
|
1765
|
+
+ "its own retention binding (storage limitation's evidence) and impersonalized on erasure. A host whose container binds no "
|
|
1766
|
+
+ "provider-call recorder writes neither row. The per-agent guardrail counts name each agent by its declared ref only, and an "
|
|
1767
|
+
+ "agent call with no record is counted for no agent.",
|
|
1768
|
+
},
|
|
1769
|
+
"evidence-requirement:ai-use-classification-declaration": {
|
|
1770
|
+
schemaVersion: 1,
|
|
1771
|
+
applicability: "Applies to every application with an AI use in the contracted service: a capability-level use posture, or an agent running on an AI "
|
|
1772
|
+
+ "capability the service reaches. An application with none runs no AI use, and that is the answer rather than a gap. The use DOMAIN is "
|
|
1773
|
+
+ "owed everywhere; its regulatory reading is computed only where the application declares a jurisdiction the EU AI Act reaches.",
|
|
1774
|
+
clauses: [
|
|
1775
|
+
{
|
|
1776
|
+
clauseRef: "C1",
|
|
1777
|
+
obligation: "Every AI use of the contracted service declares its use domain from the closed vocabulary, and its regulatory class is derived from "
|
|
1778
|
+
+ "that domain where the EU AI Act applies.",
|
|
1779
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1780
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1781
|
+
homeRationale: "The domain is a committed declaration (`aiUse` on an agent, `aiUsePosture` on a capability), and nothing about a capability being "
|
|
1782
|
+
+ "enabled can establish it: the same model backs a summary and a candidate screening. The class is a pure function of the declared "
|
|
1783
|
+
+ "domain and the declared jurisdictions, so the commit is the evidence and the reading is re-derived from it.",
|
|
1784
|
+
},
|
|
1785
|
+
],
|
|
1786
|
+
minimumEvidence: "The AI uses of the contracted service: each capability-level posture and each agent on a contracted-service AI capability, read from the "
|
|
1787
|
+
+ "agents the application registers. For each, its declared domain or its absence, and, where the EU AI Act applies, the derived tier "
|
|
1788
|
+
+ "with the Annex III area or Article 5(1) point it rests on. The agents that declare no use, and the uses that declare no domain.",
|
|
1789
|
+
verificationSteps: [
|
|
1790
|
+
"Derive the AI capabilities the contracted service reaches, with the engine's own partition.",
|
|
1791
|
+
"Read the agents the application registers through its initialization declaration, and keep those running on one of those capabilities.",
|
|
1792
|
+
"Assert each such agent, and each capability no agent covers, declares a use with a domain; name every one that does not.",
|
|
1793
|
+
"Derive the application's AI-regulation regimes from its declared jurisdictions and, where the EU AI Act applies, the tier of every domain.",
|
|
1794
|
+
"Name every prohibited-practice candidate and every high-risk candidate.",
|
|
1795
|
+
],
|
|
1796
|
+
severityGuidance: "A use declaring a practice the EU AI Act prohibits, where it applies, is a CRITICAL: the application states it does something the law "
|
|
1797
|
+
+ "forbids outright, and it is marked, never refused. A high-risk candidate is a MAJOR: Wildo does not yet support high-risk AI systems, so "
|
|
1798
|
+
+ "the use is unsupported until a person establishes the Annex III exemption applies. An agent or capability with no declared use is a MAJOR "
|
|
1799
|
+
+ "(a notice then claims for it an exemption nobody decided). A use declared without a domain is a MINOR: nobody has classified it yet, "
|
|
1800
|
+
+ "which is an honest state. Where the EU AI Act applies, a use declaring a high-risk exemption on a domain outside Annex III is a MINOR: "
|
|
1801
|
+
+ "the exemption means nothing there, so either the domain or the exemption is wrong. A use whose domain is OTHER, or an application "
|
|
1802
|
+
+ "declaring no jurisdiction, is an OBSERVATION: the class is owed.",
|
|
1803
|
+
limits: "It evidences what the application DECLARES about each use and the class the framework derives from it, never what the model does. A "
|
|
1804
|
+
+ "candidate is not a legal conclusion: the Annex III exemption and the Article 5 boundary are assessments a person makes. It reads only the "
|
|
1805
|
+
+ "agents the application registers; an engine default agent, and a tool an agent names that the application does not register, are outside it.",
|
|
1806
|
+
},
|
|
809
1807
|
"evidence-requirement:third-party-ai-component-declaration": {
|
|
810
1808
|
schemaVersion: 1,
|
|
811
1809
|
applicability: "Applies to every application with an AI capability enabled in the contracted service. An application that enables none uses no "
|
|
@@ -814,11 +1812,12 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
814
1812
|
clauses: [
|
|
815
1813
|
{
|
|
816
1814
|
clauseRef: "C1",
|
|
817
|
-
obligation: "Every AI capability enabled in the contracted service declares what it is used for and whether it decides anything about a person.",
|
|
1815
|
+
obligation: "Every AI capability enabled in the contracted service declares what it is used for and whether it decides anything about a person, at the capability or on every agent running on it.",
|
|
818
1816
|
home: AssuranceEvidenceHome.DECLARED,
|
|
819
1817
|
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
820
|
-
homeRationale: "The use posture is a committed declaration (`aiUsePosture`
|
|
821
|
-
+ "backs a summary that decides nothing and a screening that decides a great deal. The committed
|
|
1818
|
+
homeRationale: "The use posture is a committed declaration (`aiUsePosture` on a capability, `aiUse` on an agent), and a capability being enabled "
|
|
1819
|
+
+ "cannot establish it: the same model backs a summary that decides nothing and a screening that decides a great deal. The committed "
|
|
1820
|
+
+ "configuration and agent definitions are the evidence.",
|
|
822
1821
|
},
|
|
823
1822
|
{
|
|
824
1823
|
clauseRef: "C2",
|
|
@@ -830,11 +1829,12 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
830
1829
|
},
|
|
831
1830
|
],
|
|
832
1831
|
minimumEvidence: "C1: the enabled capabilities intersected with the AI capabilities of the contracted service, and for each its declared posture — the "
|
|
833
|
-
+ "consequence and the purpose, and the human-review arrangement where the AI decides or assists a decision
|
|
1832
|
+
+ "consequence and the purpose, and the human-review arrangement where the AI decides or assists a decision — declared at the capability or "
|
|
1833
|
+
+ "on every agent running on it. C2: for each runtime provider "
|
|
834
1834
|
+ "selected for such a capability, its compliance choice and, when vendor-hosted, its processor-register entry.",
|
|
835
1835
|
verificationSteps: [
|
|
836
1836
|
"Enumerate the AI capabilities enabled in the contracted service, derived from the framework's own vocabulary rather than a list.",
|
|
837
|
-
"Assert each carries a declared use posture; name every one that does not.",
|
|
1837
|
+
"Assert each carries a declared use posture, at the capability or on every agent running on it; name every one that does not.",
|
|
838
1838
|
"Enumerate the runtime providers selected for those capabilities, excluding the build-time companion scope.",
|
|
839
1839
|
"Assert each is SELF_HOSTED or appears on the processor register; name every one that does not.",
|
|
840
1840
|
],
|
|
@@ -845,6 +1845,142 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
845
1845
|
+ "a feature that in fact decides something is not caught here. It says nothing about the vendor's own training or retention terms, and "
|
|
846
1846
|
+ "nothing about an organisation's AI risk policies, which are the operator's acts rather than the product's.",
|
|
847
1847
|
},
|
|
1848
|
+
"evidence-requirement:agent-untrusted-content-screening-record": {
|
|
1849
|
+
schemaVersion: 1,
|
|
1850
|
+
applicability: "Applies to every application, and within it to every agent that declares prompt-injection screening, whether or not a "
|
|
1851
|
+
+ "content-safety classifier is selected: with none selected, C4's refusal is the control. An agent that declares no screening is "
|
|
1852
|
+
+ "outside the population: whether it should screen what it reads is the operator's judgement about where its tools reach, and C1 "
|
|
1853
|
+
+ "records that judgement rather than second-guessing it. An application where no agent declares screening answers with that.",
|
|
1854
|
+
clauses: [
|
|
1855
|
+
{
|
|
1856
|
+
clauseRef: "C1",
|
|
1857
|
+
obligation: "Whether an agent's third-party content is screened is a DECLARED decision, per agent.",
|
|
1858
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1859
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1860
|
+
homeRationale: "The screening is the agent's committed `contentGuardrails.promptInjection` declaration, and nothing at runtime turns it on or off. "
|
|
1861
|
+
+ "Which agents read third-party content is also in the commit — their declared tools and the inbound deliveries that start their "
|
|
1862
|
+
+ "turns — so the population owing a decision and the decision itself are both read from it.",
|
|
1863
|
+
},
|
|
1864
|
+
{
|
|
1865
|
+
clauseRef: "C2",
|
|
1866
|
+
obligation: "Content judged an attack, and content that could not be screened, never reaches the agent's model.",
|
|
1867
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1868
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1869
|
+
homeRationale: "The fail-closed half. A healthy deployment never shows it: with a classifier that answers \"clean\", a withholding and a "
|
|
1870
|
+
+ "pass-through implementation behave identically. The condition has to be created — a classifier double flagging a tool result, and "
|
|
1871
|
+
+ "one that does not answer — and what the model then received asserted.",
|
|
1872
|
+
},
|
|
1873
|
+
{
|
|
1874
|
+
clauseRef: "C3",
|
|
1875
|
+
obligation: "Each withheld piece of content is attributable in the audit trail, with where it came from.",
|
|
1876
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1877
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1878
|
+
homeRationale: "A durable row class in the engine's own event store, so one collector query serves every application. A detected attack and a "
|
|
1879
|
+
+ "screening that did not run are separate members on purpose: the first is a conclusion about a real attempt to steer the agent, "
|
|
1880
|
+
+ "the second an operational fault, and merging them would bury the alert that matters.",
|
|
1881
|
+
emittedCountLimit: { unreached: "a count shows withholdings were recorded, not that every withholding was: whether each names its source is read from the distribution, and a row-less withholding is C2's question" },
|
|
1882
|
+
requiredAuditEventMembers: [CoreAuditableEventType.AGENT_CONTENT_QUARANTINED, CoreAuditableEventType.AGENT_CONTENT_SCREENING_FAILED],
|
|
1883
|
+
reportedAuditEventDataDistribution: { field: "contentSource" },
|
|
1884
|
+
},
|
|
1885
|
+
{
|
|
1886
|
+
clauseRef: "C4",
|
|
1887
|
+
obligation: "Where no classifier can serve an agent's screening, a turn that would bring it third-party content — one offering it tools, or one an inbound delivery would start — is refused.",
|
|
1888
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1889
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1890
|
+
homeRationale: "The configuration half of failing closed, and like C2 not reachable on a conformant deployment: with a classifier selected the "
|
|
1891
|
+
+ "refusal never fires. It is created by starting a turn that offers a screening agent tools, and by sending it an inbound "
|
|
1892
|
+
+ "delivery, with no classifier selected (or one that does not declare the screening operation), and asserting each is refused "
|
|
1893
|
+
+ "rather than run unscreened. A buffered call, which offers the agent no tools, reads no third-party content, so screening owes "
|
|
1894
|
+
+ "it nothing and it is not refused.",
|
|
1895
|
+
},
|
|
1896
|
+
],
|
|
1897
|
+
minimumEvidence: "C1: for each agent, its declared screening scope, beside its declared tools and the inbound deliveries that start its turns. C2: a driven "
|
|
1898
|
+
+ "turn whose tool result a classifier double flags, and one whose classifier does not answer, each showing the model received the "
|
|
1899
|
+
+ "withheld notice rather than the content; and a driven inbound delivery judged an attack that started no turn. C3: the two audit members "
|
|
1900
|
+
+ "over the window, with the distribution of `contentSource`. C4: a driven tool-offering turn, and a driven inbound delivery, for a screening agent with no usable classifier, each refused.",
|
|
1901
|
+
verificationSteps: [
|
|
1902
|
+
"Enumerate the agents and read each one's declared screening scope; list those whose tools or inbound deliveries bring in other people's content and declare none.",
|
|
1903
|
+
"Drive a turn whose tool result the classifier flags, and assert the model's next input carries the withheld notice and not the result.",
|
|
1904
|
+
"Drive the same turn with the classifier unreachable, and assert the same notice — an unanswered screening must not release content.",
|
|
1905
|
+
"Drive an inbound delivery the classifier flags, and assert it starts no turn.",
|
|
1906
|
+
"Query the two audit members over the window, and read the `contentSource` distribution beside the count.",
|
|
1907
|
+
"With no classifier selected, start a turn offering a screening agent tools and send it an inbound delivery, and assert each is refused before any model call.",
|
|
1908
|
+
],
|
|
1909
|
+
severityGuidance: "Content judged an attack that reached the model is a MAJOR: the agent can act on a stranger's instructions, which is the harm the "
|
|
1910
|
+
+ "objective names. Content that could not be screened and reached the model anyway is a MAJOR for the same reason. An agent that "
|
|
1911
|
+
+ "declares screening and ran with no classifier is a MAJOR. An agent whose tools read the web or third-party records and that declares "
|
|
1912
|
+
+ "no screening is an OBSERVATION, not a nonconformity: the decision is the operator's, and C1 records it. A withholding with no audit "
|
|
1913
|
+
+ "row is an OBSERVATION where the content was withheld and a MAJOR where it was not.",
|
|
1914
|
+
limits: "It evidences that screening runs where it is declared and fails closed, never how well the classifier detects: a missed attack is the "
|
|
1915
|
+
+ "vendor's detection quality, and no product declaration can attest it. The content a person driving their own conversation writes is "
|
|
1916
|
+
+ "not screened by design (it is their own instruction), and text the application itself composes into an agent's input is the "
|
|
1917
|
+
+ "application's. The classifier is a recipient of the model-side text and is evidenced as a processor under third-party processing, not here.",
|
|
1918
|
+
},
|
|
1919
|
+
"evidence-requirement:agent-output-moderation-record": {
|
|
1920
|
+
schemaVersion: 1,
|
|
1921
|
+
applicability: "Applies to every application, and within it to every agent that declares an output-moderation policy, whether or not a "
|
|
1922
|
+
+ "content-safety classifier is selected: with none selected, C4's refusal is the control. An agent that declares none is outside "
|
|
1923
|
+
+ "the population, and C1 records that decision.",
|
|
1924
|
+
clauses: [
|
|
1925
|
+
{
|
|
1926
|
+
clauseRef: "C1",
|
|
1927
|
+
obligation: "Whether, and at what severity per harm category, an agent's output is withheld is a DECLARED decision, per agent.",
|
|
1928
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
1929
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
1930
|
+
homeRationale: "The policy is the agent's committed `contentGuardrails.outputModeration.blockAtOrAbove`, a severity per category, and a category "
|
|
1931
|
+
+ "absent from it is not judged. The commit is therefore the evidence of what is judged and where the line sits.",
|
|
1932
|
+
},
|
|
1933
|
+
{
|
|
1934
|
+
clauseRef: "C2",
|
|
1935
|
+
obligation: "Output at or above the policy's severity, and output that could not be judged, is not delivered: not shown, not stored, not sent, "
|
|
1936
|
+
+ "and a tool call written in it does not run.",
|
|
1937
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1938
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1939
|
+
homeRationale: "The withholding sits below the model SDK, so the question is about every kind of output at once — a reply, shown reasoning, a "
|
|
1940
|
+
+ "tool call's input, a structured answer — and a row cannot show that a tool did NOT run. A classifier double flagging each kind, "
|
|
1941
|
+
+ "and one that does not answer, has to be driven and what left the agent asserted.",
|
|
1942
|
+
},
|
|
1943
|
+
{
|
|
1944
|
+
clauseRef: "C3",
|
|
1945
|
+
obligation: "Each withheld output is attributable in the audit trail, with its kind and the categories it reached, never its text.",
|
|
1946
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
1947
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
1948
|
+
homeRationale: "A durable row class in the engine's own event store. Output beyond the policy and output that could not be judged are separate "
|
|
1949
|
+
+ "members: one says the control held against something the model wrote, the other that it did not run.",
|
|
1950
|
+
emittedCountLimit: { unreached: "a count shows withholdings were recorded, not that every withholding was: the kind is read from the distribution, the categories are opened per row, and a row-less withholding is C2's question" },
|
|
1951
|
+
requiredAuditEventMembers: [CoreAuditableEventType.AGENT_OUTPUT_WITHHELD, CoreAuditableEventType.AGENT_OUTPUT_MODERATION_FAILED],
|
|
1952
|
+
reportedAuditEventDataDistribution: { field: "outputKind" },
|
|
1953
|
+
},
|
|
1954
|
+
{
|
|
1955
|
+
clauseRef: "C4",
|
|
1956
|
+
obligation: "An agent whose policy no selected classifier can serve, in every category it names, is refused before its model is called.",
|
|
1957
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
1958
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
1959
|
+
homeRationale: "Not reachable on a conformant deployment, so it is created: a policy naming a category the selected classifier does not assess, "
|
|
1960
|
+
+ "invoked, and refused rather than run with that category silently unjudged.",
|
|
1961
|
+
},
|
|
1962
|
+
],
|
|
1963
|
+
minimumEvidence: "C1: for each agent, its declared policy — the categories judged and the severity each is withheld at. C2: driven turns in which a "
|
|
1964
|
+
+ "classifier double flags a reply, a tool call's input and a structured answer, and one in which it does not answer, each showing "
|
|
1965
|
+
+ "nothing was delivered and the tool did not run. C3: the two audit members over the window, with the distribution of `outputKind`. "
|
|
1966
|
+
+ "C4: a driven invocation whose policy names a category the classifier does not assess, refused.",
|
|
1967
|
+
verificationSteps: [
|
|
1968
|
+
"Enumerate the agents and read each one's declared moderation policy.",
|
|
1969
|
+
"Drive a turn whose reply the classifier flags at the policy's severity, and assert nothing from the flagged chunk on was shown, and no reply was stored or returned.",
|
|
1970
|
+
"Drive a turn in which the model writes a tool call the classifier flags, and assert the tool did not run and no approver was asked.",
|
|
1971
|
+
"Drive a structured answer the classifier flags, and assert the flow step received no output.",
|
|
1972
|
+
"Drive a turn with the classifier unreachable, and assert the same withholding — an unjudged output must not be delivered.",
|
|
1973
|
+
"Query the two audit members over the window, and read the `outputKind` distribution beside the count.",
|
|
1974
|
+
"Invoke an agent whose policy names a category the classifier does not assess, and assert it is refused before any model call.",
|
|
1975
|
+
],
|
|
1976
|
+
severityGuidance: "Output beyond the declared policy that was delivered is a MAJOR: the agent said or did what the operator declared it must not. An "
|
|
1977
|
+
+ "unjudged output that was delivered is a MAJOR. A tool that ran on a flagged call is a MAJOR even where the reply was withheld, because "
|
|
1978
|
+
+ "the act already left the system. An agent with no policy is an OBSERVATION: the decision is the operator's, and C1 records it.",
|
|
1979
|
+
limits: "It evidences that moderation runs where it is declared and fails closed, never the classifier's accuracy or that its categories cover "
|
|
1980
|
+
+ "every harm: a category the selected vendor does not assess cannot be judged, and an agent declaring one is refused (C4); misinformation is not a harm category at all. Text streamed "
|
|
1981
|
+
+ "to a person is released in judged blocks, so what a person saw before a later block was withheld was itself judged; nothing here "
|
|
1982
|
+
+ "evidences what a person did with it.",
|
|
1983
|
+
},
|
|
848
1984
|
"evidence-requirement:multi-factor-enrolment-record": {
|
|
849
1985
|
schemaVersion: 1,
|
|
850
1986
|
applicability: "Applies to every application: the verification authority is engine-owned and unconditional. Note the demand is "
|
|
@@ -1080,7 +2216,9 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
1080
2216
|
+ "reached the audit pipeline\" (slug `audit_observation_dropped`), whose context carries the lost action's identity, "
|
|
1081
2217
|
+ "and the two counters `audit_observation_dropped_total` (the observer never reached the pipeline) and "
|
|
1082
2218
|
+ "`audit_store_failure_total` (the pipeline reached the store and the store refused). "
|
|
1083
|
-
+ "Both counters are needed because they cover
|
|
2219
|
+
+ "Both counters are needed because they mostly cover different loss shapes. They overlap where the audit method "
|
|
2220
|
+
+ "records with `SURFACE_TO_CALLER`: a store refusal there is counted by both, as `emission_rejected` (or "
|
|
2221
|
+
+ "`emission_timed_out`, when the caller stopped waiting first) beside the store failure. So never sum the two. "
|
|
1084
2222
|
+ "The boot diagnostic for an absent delivery path and the dead-letter drain may also be gathered, but they evidence "
|
|
1085
2223
|
+ "the PIPELINE's health rather than the attribution this clause demands, and must not be substituted for it.",
|
|
1086
2224
|
verificationSteps: [
|
|
@@ -1315,6 +2453,320 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
1315
2453
|
+ "indistinguishable from a well-run approval. What it does buy is that a single compromised or mistaken operator "
|
|
1316
2454
|
+ "cannot reach application-wide data alone.",
|
|
1317
2455
|
},
|
|
2456
|
+
/*
|
|
2457
|
+
* #2488 — maker-checker on a decision the product holds for review. The decision space is
|
|
2458
|
+
* `HumanDecision_RefusalReason` (named whole by `capability:pending-decision-authorization`); the record is the
|
|
2459
|
+
* refusal row each door writes with that reason.
|
|
2460
|
+
*/
|
|
2461
|
+
"evidence-requirement:pending-decision-refusal-record": {
|
|
2462
|
+
schemaVersion: 1,
|
|
2463
|
+
applicability: "Applies to every application: the human-decision planes and their one authorization predicate are engine code, with "
|
|
2464
|
+
+ "nothing to enable. An application whose agents never raise a decision for review is in scope and has no rows; that is "
|
|
2465
|
+
+ "a quiet window, not a gap. A person's operation is in it when its application declares `requiresApproval` on it (#2288): "
|
|
2466
|
+
+ "the request it holds is answered through the same predicate and refused with the same row. One declared without it "
|
|
2467
|
+
+ "is applied at once and is outside it.",
|
|
2468
|
+
clauses: [
|
|
2469
|
+
{
|
|
2470
|
+
clauseRef: "C1",
|
|
2471
|
+
obligation: "Under any route other than the requester's own, the person whose work raised a decision can never act on it, and an "
|
|
2472
|
+
+ "attempt is recorded as the separation refusing them.",
|
|
2473
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
2474
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
2475
|
+
homeRationale: "Every door that refuses an act on a pending decision writes HUMAN_DECISION_ANSWER_REFUSED with the predicate's own "
|
|
2476
|
+
+ "reason, so the separation has a row of its own: the reason REQUESTER_MAY_NOT_ACT. Counting the event type alone "
|
|
2477
|
+
+ "would let every ordinary refusal — a reviewer without the role, a stale client — read as the separation holding.",
|
|
2478
|
+
emittedCountLimit: {
|
|
2479
|
+
unreached: "that no requester ever ACTED on their own decision — a claim about rows that do not exist, which a count of refusals "
|
|
2480
|
+
+ "cannot reach; a HUMAN_DECISION_RESOLVED row whose decidedByUserId is the decision's requester is what would refute it",
|
|
2481
|
+
},
|
|
2482
|
+
requiredAuditEventMembers: [CoreAuditableEventType.HUMAN_DECISION_ANSWER_REFUSED],
|
|
2483
|
+
requiredAuditEventDataMatch: {
|
|
2484
|
+
field: "reason",
|
|
2485
|
+
values: [HumanDecision_RefusalReason.REQUESTER_MAY_NOT_ACT],
|
|
2486
|
+
},
|
|
2487
|
+
},
|
|
2488
|
+
{
|
|
2489
|
+
clauseRef: "C2",
|
|
2490
|
+
obligation: "Every refused act on a pending decision is recorded with the reason the predicate gave.",
|
|
2491
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
2492
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
2493
|
+
homeRationale: "The same rows as C1, read whole. The reason distribution is reported rather than matched because every reason is "
|
|
2494
|
+
+ "evidence: ELIGIBLE_ROLE_NOT_HELD is the role re-checked at the moment of acting, NOT_THE_ASSIGNEE the route holding, "
|
|
2495
|
+
+ "and a reviewer reads the mix, not a total.",
|
|
2496
|
+
emittedCountLimit: {
|
|
2497
|
+
unreached: "that every refused act produced a row — the doors emit fire-and-forget, so a lost row and an act nobody attempted read "
|
|
2498
|
+
+ "the same in the trail",
|
|
2499
|
+
},
|
|
2500
|
+
requiredAuditEventMembers: [CoreAuditableEventType.HUMAN_DECISION_ANSWER_REFUSED],
|
|
2501
|
+
reportedAuditEventDataDistribution: { field: "reason" },
|
|
2502
|
+
},
|
|
2503
|
+
{
|
|
2504
|
+
clauseRef: "C3",
|
|
2505
|
+
obligation: "The one predicate refuses the requester every act — answer, claim, reassign, escalate, pause — under a non-owner route, "
|
|
2506
|
+
+ "re-checks the eligible role when the person acts, and a deadline rule can hand a decision only to someone it would "
|
|
2507
|
+
+ "admit to answer it.",
|
|
2508
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
2509
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
2510
|
+
homeRationale: "A property of the predicate over every act and route, which rows sample only where somebody happened to try. The "
|
|
2511
|
+
+ "predicate is pure and pinned as a truth table, and the deadline reassignment is driven against it, so the evidence is "
|
|
2512
|
+
+ "a driven result: run the tests the evidence markers name.",
|
|
2513
|
+
},
|
|
2514
|
+
],
|
|
2515
|
+
minimumEvidence: "C1: HUMAN_DECISION_ANSWER_REFUSED rows in the window whose `reason` is REQUESTER_MAY_NOT_ACT, and none required — an empty "
|
|
2516
|
+
+ "set is a quiet window. C2: every HUMAN_DECISION_ANSWER_REFUSED row in the window, with the distribution of `reason`. C3: the "
|
|
2517
|
+
+ "driven results of the predicate's truth table and of the deadline reassignment.",
|
|
2518
|
+
verificationSteps: [
|
|
2519
|
+
"Gather HUMAN_DECISION_ANSWER_REFUSED rows for the window and report the distribution of `reason` (C2).",
|
|
2520
|
+
"Count the rows whose `reason` is REQUESTER_MAY_NOT_ACT (C1); do not count other reasons toward it.",
|
|
2521
|
+
"Gather HUMAN_DECISION_RESOLVED rows for the window and, for each decision on a non-owner route, compare decidedByUserId "
|
|
2522
|
+
+ "with the decision's requester. A match is the finding this demand exists for.",
|
|
2523
|
+
"Run the tests the evidence markers name for C3 and record their result.",
|
|
2524
|
+
],
|
|
2525
|
+
severityGuidance: "A resolved decision on a non-owner route whose decider is its requester is a MAJOR, and the only finding here that "
|
|
2526
|
+
+ "invalidates the control rather than the evidence. A failing predicate or deadline test is a MAJOR against the framework. "
|
|
2527
|
+
+ "Zero refusal rows is not a finding: an application whose requesters never tried has nothing to record.",
|
|
2528
|
+
limits: "It evidences that two ACCOUNTS were involved, never two people: a requester holding a second account with the eligible role "
|
|
2529
|
+
+ "defeats it and is indistinguishable in the trail. It covers the decisions the human-decision planes raise, held operations "
|
|
2530
|
+
+ "included (#2288); an operation its application did not declare `requiresApproval` on is applied at once, so nothing "
|
|
2531
|
+
+ "here evidences maker-checker on it.",
|
|
2532
|
+
},
|
|
2533
|
+
/*
|
|
2534
|
+
* #2488 — enforcement of consent, the half `consent-decision-transition-record` disclaims. The decision space is
|
|
2535
|
+
* `ConsentGateRefusal` (named whole by `capability:consent-gate-enforcement`); the record is the denial row.
|
|
2536
|
+
*/
|
|
2537
|
+
"evidence-requirement:consent-gate-refusal-record": {
|
|
2538
|
+
schemaVersion: 1,
|
|
2539
|
+
applicability: "Applies to every application. One that requires no consent purpose of its service, declares none on an operation and runs "
|
|
2540
|
+
+ "no consent-based browser provider withholds nothing and is in scope only to confirm that; that is the answer, not a gap.",
|
|
2541
|
+
clauses: [
|
|
2542
|
+
{
|
|
2543
|
+
clauseRef: "C1",
|
|
2544
|
+
obligation: "A person who has not granted a purpose the service requires, or that the operation they reached declares it cannot run "
|
|
2545
|
+
+ "without, is refused, and the refusal is recorded.",
|
|
2546
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
2547
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
2548
|
+
homeRationale: "The gate records its refusal like every other authorization refusal: RESOURCE_OPERATION_DENIED, with the two consent "
|
|
2549
|
+
+ "reasons distinguishing the whole service withheld from one capability withheld. The type alone over-counts — it is the "
|
|
2550
|
+
+ "row for every refusal in the product — so only those two reasons are this clause's evidence.",
|
|
2551
|
+
emittedCountLimit: {
|
|
2552
|
+
unreached: "that nobody without the purpose was SERVED — a claim about requests that were not refused; and the denial emit is "
|
|
2553
|
+
+ "best-effort, so a refusal whose row was lost reads like no attempt at all",
|
|
2554
|
+
},
|
|
2555
|
+
requiredAuditEventMembers: [CoreAuditableEventType.RESOURCE_OPERATION_DENIED],
|
|
2556
|
+
requiredAuditEventDataMatch: {
|
|
2557
|
+
field: "denialReason",
|
|
2558
|
+
values: [ResourceOperationDenialReason.SERVICE_CONSENT_REQUIRED, ResourceOperationDenialReason.CAPABILITY_CONSENT_REQUIRED],
|
|
2559
|
+
},
|
|
2560
|
+
},
|
|
2561
|
+
{
|
|
2562
|
+
clauseRef: "C2",
|
|
2563
|
+
obligation: "What the application requires of its whole service, and what each operation declares it rests on and what a missing "
|
|
2564
|
+
+ "purpose costs there, is stated in its configuration.",
|
|
2565
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
2566
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
2567
|
+
homeRationale: "The service requirement (`compliance.serviceConsentRequirement.requiredPurposes`) and each operation's `consentGate` are "
|
|
2568
|
+
+ "committed configuration, and the cost of a missing purpose is a member of `ConsentGateRefusal`. Rows cannot show what "
|
|
2569
|
+
+ "was required where nobody was refused, so the declaration is read.",
|
|
2570
|
+
},
|
|
2571
|
+
{
|
|
2572
|
+
clauseRef: "C3",
|
|
2573
|
+
obligation: "The requirement is enforced on every user-initiated surface — REST, MCP, A2A and conversational turns, started or "
|
|
2574
|
+
+ "resumed — and a consent that cannot be read refuses the request, as unavailable rather than as a consent the "
|
|
2575
|
+
+ "person must grant.",
|
|
2576
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
2577
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
2578
|
+
homeRationale: "Coverage of SURFACES and the fail-closed read are properties of the code path, which a production window samples only "
|
|
2579
|
+
+ "where a refused person happened to arrive. Driven: the gate and the conversational admission are exercised by the tests "
|
|
2580
|
+
+ "the evidence markers name.",
|
|
2581
|
+
},
|
|
2582
|
+
{
|
|
2583
|
+
clauseRef: "C4",
|
|
2584
|
+
obligation: "A browser-side provider resting on a consent purpose is started only once that purpose is granted and only if a "
|
|
2585
|
+
+ "withdrawal can stop it; one that would inject a script nothing could stop is never started.",
|
|
2586
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
2587
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
2588
|
+
homeRationale: "The browser keeps no record: a withheld provider is a console diagnostic, which no store holds. The same rule is applied at "
|
|
2589
|
+
+ "the provider package's companion barrel, at hydration, by the SDK runner and by the script collector, and each is "
|
|
2590
|
+
+ "driven by the tests the evidence markers name.",
|
|
2591
|
+
},
|
|
2592
|
+
],
|
|
2593
|
+
minimumEvidence: "C1: RESOURCE_OPERATION_DENIED rows in the window whose `denialReason` is SERVICE_CONSENT_REQUIRED or "
|
|
2594
|
+
+ "CAPABILITY_CONSENT_REQUIRED. C2: the declared service requirement and every operation's consent declaration. C3 and C4: the "
|
|
2595
|
+
+ "driven results of the gate, the conversational admission and the four browser-provider applications of the stop rule.",
|
|
2596
|
+
verificationSteps: [
|
|
2597
|
+
"Read the declared service requirement and enumerate every operation declaring a consent gate, with the refusal each declares (C2).",
|
|
2598
|
+
"Gather RESOURCE_OPERATION_DENIED rows for the window and count those with the two consent reasons, per reason (C1).",
|
|
2599
|
+
"Run the tests the evidence markers name for C3 and C4 and record their result.",
|
|
2600
|
+
],
|
|
2601
|
+
severityGuidance: "A failing gate or provider test is a MAJOR against the framework: a person's withheld consent is then not honoured. A declared "
|
|
2602
|
+
+ "requirement with no refusal rows is not a finding on its own — everybody may have granted it. A declaration whose refusal is "
|
|
2603
|
+
+ "INFORM refuses nothing by design and is not a gap.",
|
|
2604
|
+
limits: "It evidences that processing the product controls is refused or not started, never what a third party does with data it "
|
|
2605
|
+
+ "already received before a withdrawal. It says nothing about whether consent is the right basis for a purpose, which is "
|
|
2606
|
+
+ "the lawful-basis demand's question, and nothing about how the decision was recorded, which is the consent-record demand's.",
|
|
2607
|
+
},
|
|
2608
|
+
/*
|
|
2609
|
+
* #2538 — approval before presentation: the governed-document approval gate, armed at Gate 3. The decision space is the
|
|
2610
|
+
* switch and the approval state it reads (named whole by `capability:governed-document-approval-gate`); the record is the
|
|
2611
|
+
* interaction row, which names the approval state of the release a person met.
|
|
2612
|
+
*/
|
|
2613
|
+
"evidence-requirement:governed-document-presentation-record": {
|
|
2614
|
+
schemaVersion: 1,
|
|
2615
|
+
applicability: "Applies to every application: the gate and the interaction record are engine code. One that has not armed the gate is in "
|
|
2616
|
+
+ "scope and does not meet C2 — it shows and asks for documents nobody approved, which is what arming the gate at Gate 3 ends. "
|
|
2617
|
+
+ "One that publishes no governed document shows and asks for nothing; that is the answer, not a gap.",
|
|
2618
|
+
clauses: [
|
|
2619
|
+
{
|
|
2620
|
+
clauseRef: "C1",
|
|
2621
|
+
obligation: "Every act a person performs with a governed document is recorded against the release they met and names whether that "
|
|
2622
|
+
+ "release was approved; once the gate is armed, no such record names an unapproved release.",
|
|
2623
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
2624
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
2625
|
+
homeRationale: "GOVERNED_DOCUMENT_INTERACTION_RECORDED is written for every notice delivery, acceptance and consent decision the product "
|
|
2626
|
+
+ "records, and `interaction.releaseApprovalState` is the approval state of that release at that instant. The distribution "
|
|
2627
|
+
+ "is the evidence: with the gate armed, an UNAPPROVED row is a person who met or accepted a document nobody approved — "
|
|
2628
|
+
+ "through a page loaded before the gate was armed, or a client still holding a release it is no longer sent.",
|
|
2629
|
+
emittedCountLimit: {
|
|
2630
|
+
unreached: "that nobody was SHOWN an unapproved document: a page view leaves a row only for a release permitting NOTICE_DELIVERED, "
|
|
2631
|
+
+ "and the emit is best-effort, so a lost row reads like no visit at all",
|
|
2632
|
+
},
|
|
2633
|
+
requiredAuditEventMembers: [CoreAuditableEventType.GOVERNED_DOCUMENT_INTERACTION_RECORDED],
|
|
2634
|
+
reportedAuditEventDataDistribution: { field: "interaction.releaseApprovalState" },
|
|
2635
|
+
},
|
|
2636
|
+
{
|
|
2637
|
+
clauseRef: "C2",
|
|
2638
|
+
obligation: "The application arms the gate in its configuration, and every release it declares states whether it was approved.",
|
|
2639
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
2640
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
2641
|
+
homeRationale: "`compliance.governedDocumentApprovalGate` is committed configuration and defaults to UNARMED; every declared release "
|
|
2642
|
+
+ "carries `approvalState`, projected from its committed approval decisions. Rows cannot show what the gate withheld from "
|
|
2643
|
+
+ "nobody, so the declaration is read.",
|
|
2644
|
+
},
|
|
2645
|
+
{
|
|
2646
|
+
clauseRef: "C3",
|
|
2647
|
+
obligation: "Once the gate is armed, the configuration sent to a reader carries no unapproved release and no draft, the public page "
|
|
2648
|
+
+ "resolves neither, and the backend owes no acceptance of, and records no delivery against, an unapproved release.",
|
|
2649
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
2650
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
2651
|
+
homeRationale: "These are properties of code paths, which a production window samples only where an unapproved release happened to exist. "
|
|
2652
|
+
+ "Driven by the tests the evidence markers name. The registration control, the first-session gate and the consent panel's "
|
|
2653
|
+
+ "notice link read the switch as well and are not driven armed: with the gate armed the projection sends them no "
|
|
2654
|
+
+ "unapproved release, and that projection is driven here, so their own reading is a second layer rather than the control.",
|
|
2655
|
+
},
|
|
2656
|
+
],
|
|
2657
|
+
minimumEvidence: "C1: GOVERNED_DOCUMENT_INTERACTION_RECORDED rows in the window and their distribution over `interaction.releaseApprovalState`, "
|
|
2658
|
+
+ "read beside the instant the gate was armed. C2: the gate's declared position and every declared release's approval state. "
|
|
2659
|
+
+ "C3: the driven results of the projection, the public page, and the backend's owed-acceptance and notice-delivery paths.",
|
|
2660
|
+
verificationSteps: [
|
|
2661
|
+
"Read the gate's declared position, the commit that armed it, and every declared release's approval state (C2).",
|
|
2662
|
+
"Gather GOVERNED_DOCUMENT_INTERACTION_RECORDED rows for the window and tally `interaction.releaseApprovalState`, separating rows "
|
|
2663
|
+
+ "written before the gate was armed from rows written after (C1).",
|
|
2664
|
+
"Run the tests the evidence markers name for C3 and record their result.",
|
|
2665
|
+
],
|
|
2666
|
+
severityGuidance: "A failing C3 test is a MAJOR against the framework: an armed application then shows or asks for a document nobody approved. "
|
|
2667
|
+
+ "An UNAPPROVED row written after the gate was armed is a MAJOR against the application's operation; one written before is the "
|
|
2668
|
+
+ "unarmed state and not a finding. An application claiming this objective with the gate unarmed does not meet C2.",
|
|
2669
|
+
limits: "It evidences what the product withholds and records, never that an approval was sound — who approved and on what review is the "
|
|
2670
|
+
+ "release's committed approval decisions, which this demand reads only as a state. It says nothing about a document's text, and "
|
|
2671
|
+
+ "nothing about copies published outside the application, such as a marketing page or the review PDF an operator exports, "
|
|
2672
|
+
+ "which states each document's standing first.",
|
|
2673
|
+
},
|
|
2674
|
+
/*
|
|
2675
|
+
* #2488 — purpose limitation for a secondary use: a tenant's agent-work decisions exported as evaluation data. The gate is
|
|
2676
|
+
* named by `capability:secondary-use-authorization`; the records are two audit rows and the export ledger.
|
|
2677
|
+
*/
|
|
2678
|
+
"evidence-requirement:secondary-use-authorization-record": {
|
|
2679
|
+
schemaVersion: 1,
|
|
2680
|
+
applicability: "Applies to every application: the export authorization is an engine-owned organization setting and the export door is "
|
|
2681
|
+
+ "engine code. A tenant that never granted an authorization holds none, or one in NOT_GRANTED, and every export it asks for "
|
|
2682
|
+
+ "is refused; that is in scope and passes on that basis.",
|
|
2683
|
+
clauses: [
|
|
2684
|
+
{
|
|
2685
|
+
clauseRef: "C1",
|
|
2686
|
+
obligation: "Every change to an export authorization is recorded, naming the status it moved to and from and the basis it rests on.",
|
|
2687
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
2688
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
2689
|
+
homeRationale: "The administrator's create and update write EVALUATION_EXPORT_AUTHORIZATION_CHANGED after the change commits. The row "
|
|
2690
|
+
+ "materialised by a first read of the settings page records a default nobody decided and is deliberately not audited, "
|
|
2691
|
+
+ "so only decisions appear. The status distribution is reported because a grant and a revocation are different "
|
|
2692
|
+
+ "evidence.",
|
|
2693
|
+
emittedCountLimit: {
|
|
2694
|
+
unreached: "that each grant names the basis it rests on and each change the status it left — fields of the row, which a count "
|
|
2695
|
+
+ "never opens",
|
|
2696
|
+
},
|
|
2697
|
+
requiredAuditEventMembers: [CoreAuditableEventType.EVALUATION_EXPORT_AUTHORIZATION_CHANGED],
|
|
2698
|
+
reportedAuditEventDataDistribution: { field: "status" },
|
|
2699
|
+
},
|
|
2700
|
+
{
|
|
2701
|
+
clauseRef: "C2",
|
|
2702
|
+
obligation: "An export asked for without an ACTIVE authorization naming its basis is refused, and the refusal is recorded.",
|
|
2703
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
2704
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
2705
|
+
homeRationale: "The export door writes EVALUATION_DATASET_EXPORT_REFUSED with why — no authorization, or a revoked one — before it "
|
|
2706
|
+
+ "refuses. Both reasons are evidence, so their distribution is reported rather than one matched.",
|
|
2707
|
+
emittedCountLimit: {
|
|
2708
|
+
unreached: "that no export LEFT without an active authorization — a claim about exports that happened, which the ledger answers "
|
|
2709
|
+
+ "(C3), not the refusals",
|
|
2710
|
+
},
|
|
2711
|
+
requiredAuditEventMembers: [CoreAuditableEventType.EVALUATION_DATASET_EXPORT_REFUSED],
|
|
2712
|
+
reportedAuditEventDataDistribution: { field: "reason" },
|
|
2713
|
+
},
|
|
2714
|
+
{
|
|
2715
|
+
clauseRef: "C3",
|
|
2716
|
+
obligation: "Every export that left names the authorization and the basis it left under, and every dataset a withdrawn basis or an "
|
|
2717
|
+
+ "erased case no longer covers is marked with why.",
|
|
2718
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
2719
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
2720
|
+
homeRationale: "The export ledger (`evaluationDatasetExports`) is an engine-owned store kept for the AUDIT_LOGS period: each row carries "
|
|
2721
|
+
+ "the authorization, the basis, the cases and case subjects it carried, and, once marked, `invalidatedAt` with an "
|
|
2722
|
+
+ "`EvaluationDatasetInvalidationReason`. The marking is a ledger update, not an audit row, so no audit member answers it.",
|
|
2723
|
+
},
|
|
2724
|
+
{
|
|
2725
|
+
clauseRef: "C4",
|
|
2726
|
+
obligation: "An authorization is never left active without its basis, a granted one is withdrawn only by revoking it, one never "
|
|
2727
|
+
+ "granted cannot be revoked, and a case subject is excluded only if a case can name it.",
|
|
2728
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
2729
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
2730
|
+
homeRationale: "Write rules refused before anything is stored, so a refused write leaves no row anywhere. Driven: the create and update "
|
|
2731
|
+
+ "prefixes are exercised by the tests the evidence markers name.",
|
|
2732
|
+
},
|
|
2733
|
+
{
|
|
2734
|
+
clauseRef: "C5",
|
|
2735
|
+
obligation: "A case subject the tenant excluded, and a privileged case or one holding special-category or criminal-offence data without "
|
|
2736
|
+
+ "its subject's recorded consent, leave in no export; while any case subject is excluded, a case whose subject cannot be read "
|
|
2737
|
+
+ "is left out.",
|
|
2738
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
2739
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
2740
|
+
homeRationale: "What did NOT leave cannot be read from what did: the ledger lists exported cases, never the ones the door left out. "
|
|
2741
|
+
+ "Driven: the export door is exercised by the tests the evidence markers name.",
|
|
2742
|
+
},
|
|
2743
|
+
{
|
|
2744
|
+
clauseRef: "C6",
|
|
2745
|
+
obligation: "Withdrawing the basis — a revocation, a removed subject consent, a newly walled-off case or a newly excluded case subject — or "
|
|
2746
|
+
+ "erasing a case marks every dataset already exported under it.",
|
|
2747
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
2748
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
2749
|
+
homeRationale: "C3 reads the marks that exist; whether every dataset that should have been marked was is a claim about the rows that "
|
|
2750
|
+
+ "were not, so it is driven: the withdrawal and the case erasure are exercised by the tests the evidence markers name.",
|
|
2751
|
+
},
|
|
2752
|
+
],
|
|
2753
|
+
minimumEvidence: "C1: EVALUATION_EXPORT_AUTHORIZATION_CHANGED rows in the window with the distribution of `status`. C2: "
|
|
2754
|
+
+ "EVALUATION_DATASET_EXPORT_REFUSED rows with the distribution of `reason`. C3: the export ledger for the window — each row's "
|
|
2755
|
+
+ "authorization and basis, and each marked row's reason. C4–C6: the driven results of the write rules, the export door and "
|
|
2756
|
+
+ "the withdrawal and erasure paths.",
|
|
2757
|
+
verificationSteps: [
|
|
2758
|
+
"Gather EVALUATION_EXPORT_AUTHORIZATION_CHANGED rows for the window and report the distribution of `status` (C1).",
|
|
2759
|
+
"Gather EVALUATION_DATASET_EXPORT_REFUSED rows and report the distribution of `reason` (C2).",
|
|
2760
|
+
"Read the export ledger for the window and assert every row names an authorization and a basis; for each marked row, read "
|
|
2761
|
+
+ "its invalidation reason (C3).",
|
|
2762
|
+
"Run the tests the evidence markers name for C4, C5 and C6 and record their result.",
|
|
2763
|
+
],
|
|
2764
|
+
severityGuidance: "A ledger row with no authorization or no basis is a MAJOR: data left for another purpose on nothing. A failing write-rule, "
|
|
2765
|
+
+ "exclusion or marking test is a MAJOR against the framework. Zero exports and zero refusals are not findings.",
|
|
2766
|
+
limits: "It evidences that a basis was RECORDED and enforced at the door, never that the recorded basis is legally sound — a "
|
|
2767
|
+
+ "compatibility assessment or a controller's instruction is the operator's statement, which a qualified reviewer judges. "
|
|
2768
|
+
+ "Marking a dataset records that it may no longer be used; it cannot recall a copy already delivered to its destination.",
|
|
2769
|
+
},
|
|
1318
2770
|
"evidence-requirement:cross-tenant-administrative-access-record": {
|
|
1319
2771
|
schemaVersion: 1,
|
|
1320
2772
|
applicability: "Applies to every application, and most sharply to those with no platform-tier administration: the demand's strongest evidence is an empty crossing set beside a live enforcement path.",
|
|
@@ -1544,11 +2996,26 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
1544
2996
|
+ "billing adapter call that pages through a list may make several HTTP requests under one row. A storage's "
|
|
1545
2997
|
+ "readiness probe is not recorded — it carries no file and acts for no one, and a probe every few seconds would "
|
|
1546
2998
|
+ "bury the calls the trail exists for. A storage a provider package contributes records only if its code calls the "
|
|
1547
|
-
+ "same recorder. Calls to a SYSTEM OF RECORD are not C4's, and
|
|
1548
|
-
+ "
|
|
1549
|
-
+ "
|
|
1550
|
-
+ "
|
|
1551
|
-
+ "
|
|
2999
|
+
+ "same recorder. Calls to a SYSTEM OF RECORD are not C4's, and they are recorded per ACT as C4's are: a read-through "
|
|
3000
|
+
+ "binding leaves one `HTTP_API_READ_THROUGH_ACCESS` row per OPERATION (a paginated list and its count are one access; "
|
|
3001
|
+
+ "a write's read-back of the row it wrote is a row of its own), a write one `HTTP_API_WRITE_THROUGH` row per attempt, "
|
|
3002
|
+
+ "a declared action one `SYSTEM_OF_RECORD_ACTION_EXECUTED` or `_REFUSED` row per attempt, a pipeline one run row per "
|
|
3003
|
+
+ "inbound run that reaches the end of what it reads (a pass spanning several runs gives each its own row) and per "
|
|
3004
|
+
+ "keyed re-read (an OUTBOUND run writes none), and one per send, and a forwarded subject erasure one "
|
|
3005
|
+
+ "`HTTP_API_EXTERNAL_ERASURE` row per forward. Each row carries `physicalCallCount`, the requests the act sent, "
|
|
3006
|
+
+ "retries and a `$metadata` probe included (#1970), so the calls to an instance in a window are the sum of its rows' "
|
|
3007
|
+
+ "counts. A count is stated when the act's row is written, so an act that spans a window boundary is counted in the "
|
|
3008
|
+
+ "window its row falls in. A FAILED act counts as reliably as one that answered: when an act's requests outlive its "
|
|
3009
|
+
+ "last row (a read whose transport failed after its retries, a run that failed mid-pass, a write whose read-back "
|
|
3010
|
+
+ "failed, a two-way pass that read after its last send), it writes one closing row when it ends, carrying them, "
|
|
3011
|
+
+ "marked `failed: true` when the act failed. A read's or a write's closing row is a read-through row for its "
|
|
3012
|
+
+ "operation; a run's is written under `externalDataPipelineRunClosing`, so a count of runs never includes it. A "
|
|
3013
|
+
+ "forward that sent requests is recorded even when it deleted nothing. An act that sent nothing writes no closing row "
|
|
3014
|
+
+ "(a single-row write or an action refused once its act began still writes its own row, with a count of zero; a "
|
|
3015
|
+
+ "bulk or filter-addressed write is refused before any act begins and writes none). Not counted: a token request, a mint "
|
|
3016
|
+
+ "or a forced re-mint after a refused token. The instance's token minter sends it and its token serves every act until "
|
|
3017
|
+
+ "it expires, so it belongs to no act, wherever the vendor hosts the token endpoint (possibly on the instance itself). "
|
|
3018
|
+
+ "Nor is any row recorded before the field existed. BROWSER-side egress — error monitoring, analytics, a frontend provider's own "
|
|
1552
3019
|
+ "calls — never reaches a backend recorder, so it is OUT of scope here and is never implied covered; its consent "
|
|
1553
3020
|
+ "is a separate requirement.",
|
|
1554
3021
|
},
|
|
@@ -1795,8 +3262,9 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
1795
3262
|
},
|
|
1796
3263
|
"evidence-requirement:lawful-basis-declaration": {
|
|
1797
3264
|
schemaVersion: 1,
|
|
1798
|
-
applicability: "Applies to every application that processes personal data
|
|
1799
|
-
+ "
|
|
3265
|
+
applicability: "Applies to every application that processes personal data where the GDPR reaches a jurisdiction it declares in "
|
|
3266
|
+
+ "`operatingJurisdictions`; an application that declares none has this jurisdiction owed, never waived (#2852). The basis is "
|
|
3267
|
+
+ "the operator's declaration, so there is nothing to opt into — an application that has declared none is covered by this "
|
|
1800
3268
|
+ "demand and reported by it as not having answered.",
|
|
1801
3269
|
clauses: [
|
|
1802
3270
|
{
|
|
@@ -2035,23 +3503,37 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
2035
3503
|
+ "framework owns no log store. The logger writes to stdout or an exporter, so the archive a collector reads is "
|
|
2036
3504
|
+ "the HOST's, on the host's retention.\n\n"
|
|
2037
3505
|
+ "This clause said EMITTED until 2026-09-10, which told a collector to query an engine-owned record class that "
|
|
2038
|
-
+ "does not exist.
|
|
2039
|
-
+ "
|
|
3506
|
+
+ "does not exist. C1 above reads rows back from the audit repository, a real store, so the two halves of this "
|
|
3507
|
+
+ "demand are gathered from two different places, which is the thing a judge most needs to be told.\n\n"
|
|
2040
3508
|
+ "It is the clause that makes a record attributable across deployments: without it two environments' rows are "
|
|
2041
3509
|
+ "indistinguishable once pooled.",
|
|
2042
3510
|
},
|
|
3511
|
+
{
|
|
3512
|
+
clauseRef: "C3",
|
|
3513
|
+
obligation: "A row written while an AI agent's tool runs names the agent beside the person or principal it acted for, whichever tool "
|
|
3514
|
+
+ "plane wrote it.",
|
|
3515
|
+
home: AssuranceEvidenceHome.UNPRODUCED,
|
|
3516
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
3517
|
+
homeRationale: "The engine opens an agent frame around every tool call, and both the execution context and the audit writer adopt it, so a "
|
|
3518
|
+
+ "tool that forgets the agent still records it (#2839). Whether a row carries the agent is readable, but whether a row "
|
|
3519
|
+
+ "written by an agent LACKS it is not: a row with no agent looks like a person's own write. Only a driven tool call read back "
|
|
3520
|
+
+ "separates them, and no scenario drives one yet (#2850); UNPRODUCED until one does, like the other driven-turn clauses.",
|
|
3521
|
+
},
|
|
2043
3522
|
],
|
|
2044
3523
|
minimumEvidence: "C1: a driven trusted-system write with a known actor and tenant, and the persisted row read back carrying both "
|
|
2045
|
-
+ "unchanged. C2: log records from the window carrying application and environment identifiers."
|
|
3524
|
+
+ "unchanged. C2: log records from the window carrying application and environment identifiers. C3: a driven agent tool "
|
|
3525
|
+
+ "call, and its rows read back naming the agent and the principal.",
|
|
2046
3526
|
verificationSteps: [
|
|
2047
3527
|
"Drive a write as a trusted system on behalf of a known principal in a known tenant.",
|
|
2048
3528
|
"Read the persisted row and assert the actor and tenant are the ones supplied, not the trusted system's own.",
|
|
2049
3529
|
"Sample log records from the window and assert application and environment are present.",
|
|
3530
|
+
"Drive an agent's tool that writes, through a framework tool and a hand-written one, and assert each row names the agent and the principal.",
|
|
2050
3531
|
],
|
|
2051
3532
|
severityGuidance: "A row naming the trusted system instead of the principal is a MAJOR — it is the intent's stated harm, a record "
|
|
2052
3533
|
+ "silently losing the principal it belongs to. A tenant replaced is a MAJOR and also a tenancy defect. A log record "
|
|
2053
3534
|
+ "without application or environment is an OBSERVATION alone and a MAJOR once records from two deployments are "
|
|
2054
|
-
+ "pooled, which is the case the clause exists for."
|
|
3535
|
+
+ "pooled, which is the case the clause exists for. Once C3 is driven, an agent's write recorded as the person's own is a MAJOR: "
|
|
3536
|
+
+ "the trail then says a person did what an agent did.",
|
|
2055
3537
|
limits: "C1 attests the write path DRIVEN, not every production write; a second path that persists rows without passing "
|
|
2056
3538
|
+ "through it is invisible here. C2 evidences presence of the identifiers, never that they are correct.",
|
|
2057
3539
|
},
|
|
@@ -2103,8 +3585,10 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
2103
3585
|
"evidence-requirement:over-limit-refusal-record": {
|
|
2104
3586
|
schemaVersion: 1,
|
|
2105
3587
|
applicability: "Applies to every application, unconditionally: the bound is engine-mounted and no application opts out. Note the "
|
|
2106
|
-
+ "asymmetry a judge must carry — the intent speaks of an unauthenticated request generally, and
|
|
2107
|
-
+ "
|
|
3588
|
+
+ "asymmetry a judge must carry — the intent speaks of an unauthenticated request generally, and five marked sites each bound "
|
|
3589
|
+
+ "one unauthenticated or unverified input: the OAuth token body cap (the only one C1 reads), a submission challenge on an "
|
|
3590
|
+
+ "operation that declares one, an unverified notification's re-reads, and an inbound email over a declared limit (marked at "
|
|
3591
|
+
+ "its reception and at its audit event). An authenticated caller's burst bound is `WILDO.OPERATIONS.AUTHENTICATED_BURST_LIMIT`'s.",
|
|
2108
3592
|
clauses: [
|
|
2109
3593
|
{
|
|
2110
3594
|
clauseRef: "C1",
|
|
@@ -2513,8 +3997,9 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
2513
3997
|
},
|
|
2514
3998
|
"evidence-requirement:refusal-classification-record": {
|
|
2515
3999
|
schemaVersion: 1,
|
|
2516
|
-
applicability: "Applies to every application.
|
|
2517
|
-
+ "single sign-on is configured,
|
|
4000
|
+
applicability: "Applies to every application. The three marked controls are engine-owned. The second is reachable only where "
|
|
4001
|
+
+ "enterprise single sign-on is configured, and the third only where an inbound delivery endpoint is declared; each "
|
|
4002
|
+
+ "narrows its POPULATION and not its obligation.",
|
|
2518
4003
|
clauses: [
|
|
2519
4004
|
{
|
|
2520
4005
|
clauseRef: "C1",
|
|
@@ -2543,20 +4028,43 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
2543
4028
|
+ "configuration problem with an actionable remedy, never an assertion-invalid verdict. Both outcomes refuse; only "
|
|
2544
4029
|
+ "the classification tells an operator whether to look at the identity provider or at their own toolchain.",
|
|
2545
4030
|
},
|
|
4031
|
+
{
|
|
4032
|
+
clauseRef: "C4",
|
|
4033
|
+
obligation: "The inbound delivery door refuses an unverifiable request, and records WHICH refusal it was — a missing "
|
|
4034
|
+
+ "verification secret apart from a signature that did not verify.",
|
|
4035
|
+
home: AssuranceEvidenceHome.EMITTED,
|
|
4036
|
+
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
4037
|
+
homeRationale: "The door answers rather than throws, and it answers both a missing secret and a bad signature identically, so the "
|
|
4038
|
+
+ "sender learns nothing about the configuration. That makes the RECORD the only place the degradation is told "
|
|
4039
|
+
+ "apart from the rejection, and the record is a row per refusal — so the clause reads rows, not a probe.\n\n"
|
|
4040
|
+
+ "The reason is REPORTED as a distribution rather than matched: every reason is evidence, and a match on the two "
|
|
4041
|
+
+ "the contrast is about would drop the stale-timestamp and oversized-body refusals a reviewer also reads. Rows "
|
|
4042
|
+
+ "fold identical refusals per source and reason per minute, so a row's `refusalCount` is the number it stands for.",
|
|
4043
|
+
emittedCountLimit: {
|
|
4044
|
+
unreached: "each refusal carries the reason it was refused for, so a window of SECRET_NOT_CONFIGURED rows reads as an unconfigured endpoint rather than as an attack",
|
|
4045
|
+
},
|
|
4046
|
+
requiredAuditEventMembers: [CoreAuditableEventType.INBOUND_DELIVERY_REJECTED],
|
|
4047
|
+
reportedAuditEventDataDistribution: { field: "reason" },
|
|
4048
|
+
},
|
|
2546
4049
|
],
|
|
2547
4050
|
minimumEvidence: "C1: a driven query filter carrying an operator object, and the whole-value refusal it produced. C2 and C3: the validator "
|
|
2548
|
-
+ "driven with its dependency removed, and the error TYPE and code it produced."
|
|
4051
|
+
+ "driven with its dependency removed, and the error TYPE and code it produced. C4: the inbound refusal rows over the "
|
|
4052
|
+
+ "window, with the distribution of their reasons.",
|
|
2549
4053
|
verificationSteps: [
|
|
2550
4054
|
"Send a filter value carrying a structured operator and assert THE VALUE DID NOT REACH THE QUERY. That is the evidence, not a request-level refusal: the guard DROPS the offending key and the request proceeds on the remaining filter, so on the agent surface the caller sees a 200. Where a refusal does occur it comes from the request contract, which is a different mechanism — asserting one on a surface that normalizes rather than parses reports a MAJOR against the marked control.",
|
|
2551
4055
|
"Assert the permitted scalar set is a named predicate and that a new structured shape would require a deliberate edit to it.",
|
|
2552
4056
|
"Remove the validator dependency, drive an assertion, and assert the refusal is a configuration error carrying a remedy.",
|
|
2553
4057
|
"Assert that refusal is NOT reported as an invalid signature — the two must be separable by an operator reading only the error.",
|
|
4058
|
+
"For an application with an inbound endpoint, remove its verification secret, send a request, and assert the refusal row "
|
|
4059
|
+
+ "names SECRET_NOT_CONFIGURED rather than SIGNATURE_INVALID — an empty window says nothing, because it is also what a "
|
|
4060
|
+
+ "door that records no reason would produce.",
|
|
2554
4061
|
],
|
|
2555
4062
|
severityGuidance: "A partially sanitized filter is a MAJOR. A validator that degrades OPEN is a MAJOR and is the objective's whole subject. "
|
|
2556
4063
|
+ "A degraded validator reporting an invalid signature is also a MAJOR, and the reason is diagnostic rather than "
|
|
2557
4064
|
+ "cryptographic: it sends the operator at the identity provider while the fault is in the local toolchain, so the outage "
|
|
2558
|
-
+ "outlives the diagnosis."
|
|
2559
|
-
|
|
4065
|
+
+ "outlives the diagnosis. An inbound refusal recorded as SIGNATURE_INVALID when the secret was missing is a MAJOR for "
|
|
4066
|
+
+ "the same reason: it reads as an attack on an endpoint that was simply never configured.",
|
|
4067
|
+
limits: "Three marked sites for a principle the intent states generally — every other control that must refuse when it cannot "
|
|
2560
4068
|
+ "evaluate is outside this evidence, and a clean result here is not an application-wide fail-closed claim. C2 and C3 "
|
|
2561
4069
|
+ "attest the BUILD driven; note in particular that the validator dependency is a COMPILED artifact whose absence is "
|
|
2562
4070
|
+ "per-workspace, so a probe in one workspace says nothing about another.",
|
|
@@ -3275,11 +4783,12 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
3275
4783
|
obligation: "A security-relevant action records who performed it, and never disguises an UNATTRIBUTABLE action as an ordinary automated one.",
|
|
3276
4784
|
home: AssuranceEvidenceHome.DECLARED,
|
|
3277
4785
|
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
3278
|
-
homeRationale: "The clause is about a VOCABULARY, and that is what makes it declarative
|
|
3279
|
-
+ "
|
|
3280
|
-
+ "
|
|
3281
|
-
+ "
|
|
3282
|
-
+ "
|
|
4786
|
+
homeRationale: "The clause is about a VOCABULARY, and that is what makes it declarative: `AuditActorKind`, named whole by "
|
|
4787
|
+
+ "`capability:audit-actor-attribution`. Its members are semantically distinct — a signed-in person, a person's "
|
|
4788
|
+
+ "authority exercised by a delegated client, a machine credential, the system acting with no human principal by "
|
|
4789
|
+
+ "design, an unauthenticated caller where none was expected, and a principal that was EXPECTED and could not be "
|
|
4790
|
+
+ "resolved. Only the last is an anomaly. Collapsing it onto an ordinary value is what hid a batch-purged tenant "
|
|
4791
|
+
+ "recorded as deleted by an unknown actor — the one case worth investigating, buried inside the routine ones.",
|
|
3283
4792
|
},
|
|
3284
4793
|
{
|
|
3285
4794
|
clauseRef: "C2",
|
|
@@ -3312,23 +4821,23 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
3312
4821
|
},
|
|
3313
4822
|
{
|
|
3314
4823
|
clauseRef: "C4",
|
|
3315
|
-
obligation: "An outbound DELETION on that plane names the subject it served and the system it reached
|
|
4824
|
+
obligation: "An outbound DELETION on that plane names the subject it served and the system it reached.",
|
|
3316
4825
|
home: AssuranceEvidenceHome.EMITTED,
|
|
3317
4826
|
plane: AssuranceEvidenceExecutionPlane.PRODUCTION,
|
|
3318
|
-
homeRationale: "
|
|
3319
|
-
+ "
|
|
3320
|
-
+ "
|
|
4827
|
+
homeRationale: "One member: a forwarded erasure's row names the subject it served and the system it reached. A second member, the "
|
|
4828
|
+
+ "per-request acknowledgement of an erasure left at a system of record, was removed with the refusal it recorded "
|
|
4829
|
+
+ "(#2559): a system of record handles requests about its own records, which the erasure receipt states rather than "
|
|
4830
|
+
+ "an audit row per request.",
|
|
3321
4831
|
emittedCountLimit: {
|
|
3322
|
-
unreached: "each deletion names the subject it served and the system it reached
|
|
4832
|
+
unreached: "each deletion names the subject it served and the system it reached",
|
|
3323
4833
|
},
|
|
3324
4834
|
requiredAuditEventMembers: [
|
|
3325
4835
|
CoreAuditableEventType.HTTP_API_EXTERNAL_ERASURE,
|
|
3326
|
-
CoreAuditableEventType.HTTP_API_NON_ERASURE_ACKNOWLEDGED,
|
|
3327
4836
|
],
|
|
3328
4837
|
},
|
|
3329
4838
|
],
|
|
3330
4839
|
minimumEvidence: "C1: the sentinel vocabulary as a named closed set. C2: an enumeration of the emitters, each resolving the sentinel "
|
|
3331
|
-
+ "through the shared resolver. C3 and C4: the
|
|
4840
|
+
+ "through the shared resolver. C3 and C4: the four outbound members over the window, with the initiating-user field "
|
|
3332
4841
|
+ "present wherever the execution had a human principal. Absence of that field means a runtime acted, which is a "
|
|
3333
4842
|
+ "value rather than a gap.",
|
|
3334
4843
|
verificationSteps: [
|
|
@@ -3337,13 +4846,11 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
3337
4846
|
"Enumerate the emitters that TAKE AN ACTOR — a small minority of the service's log methods — and assert each resolves it through the shared resolver rather than choosing a literal. Enumerating all emitters is the wrong population: most carry no actor field and cannot satisfy the step, so the count would be a statement about the service's size rather than about the invariant.",
|
|
3338
4847
|
"Sample read-through rows and assert the real initiating user is present for executions that had one, and absent for machine and internal ones.",
|
|
3339
4848
|
"Assert the emission unit PER MEMBER: one row per read-family operation execution for read-through access (a list and its count are one access), one row per outbound call for provider calls, one per tool call for agent calls. A single unit asserted across all three is a MAJOR against two working emitters.",
|
|
3340
|
-
"Query the acknowledged-non-erasure member GROUPED BY resource, and read the count as the size of the accepted hole.",
|
|
3341
4849
|
],
|
|
3342
4850
|
severityGuidance: "A row carrying the anomaly sentinel is a MAJOR and is exactly what the vocabulary exists to surface: a privileged "
|
|
3343
4851
|
+ "action happened and the trail cannot say who. A read-through row missing the real initiating user where the "
|
|
3344
4852
|
+ "execution had one is a MAJOR — attribution died at the service principal, which is the failure the field exists "
|
|
3345
|
-
+ "for. An outbound deletion with no row is a MAJOR.
|
|
3346
|
-
+ "rather than per request is a MAJOR: it is the same fact made uncountable.",
|
|
4853
|
+
+ "for. An outbound deletion with no row is a MAJOR.",
|
|
3347
4854
|
limits: "C2's enumeration attests the emitter set as enumerated. C3 and C4 attest the window, and they evidence attribution "
|
|
3348
4855
|
+ "of the calls that were MADE — a call made outside the read-through plane leaves no row here. And a resolved actor "
|
|
3349
4856
|
+ "is not a verified one: the row records who the execution context said was acting, which is a different claim from "
|
|
@@ -4299,5 +5806,59 @@ export const ASSURANCE_EVIDENCE_REQUIREMENT_BODIES = {
|
|
|
4299
5806
|
+ "production deployment declared as development admits loopback. An operator may also admit private-network SIEM "
|
|
4300
5807
|
+ "collectors by declaration; a private collector reached under that declaration is the escape working, not a finding.",
|
|
4301
5808
|
},
|
|
5809
|
+
"evidence-requirement:credential-scoping-refusal": {
|
|
5810
|
+
schemaVersion: 1,
|
|
5811
|
+
applicability: "Applies to every application: every runtime refuses these configurations at startup and the CLI before it generates anything. "
|
|
5812
|
+
+ "Its population is every provider configuration the application commits, in its own configuration and in each environment's, "
|
|
5813
|
+
+ "and every credential-naming field and module-owned key in them. An application that configures no provider has an empty "
|
|
5814
|
+
+ "C1 population, which is the answer rather than a gap.",
|
|
5815
|
+
clauses: [
|
|
5816
|
+
{
|
|
5817
|
+
clauseRef: "C1",
|
|
5818
|
+
obligation: "No committed provider configuration, the application's or an environment's, authors a key its provider module owns, points "
|
|
5819
|
+
+ "a credential-naming field at a secret its provider module does not declare, or forwards into a coding agent a secret the "
|
|
5820
|
+
+ "framework or a provider declares.",
|
|
5821
|
+
home: AssuranceEvidenceHome.DECLARED,
|
|
5822
|
+
plane: AssuranceEvidenceExecutionPlane.META_APPLICATION,
|
|
5823
|
+
homeRationale: "Every field is a committed declaration, and the refusal is a pure function of it and of the declarations it is held to: the "
|
|
5824
|
+
+ "module-owned keys, the credential-naming fields as the doors read them (`readProviderConfigurationCredentialFields`), each "
|
|
5825
|
+
+ "provider's declared slots and the framework's declared secrets. A misconfigured application does not start, so no runtime "
|
|
5826
|
+
+ "row exists to count; the configuration is the evidence. A provider whose module is not readable is reported as unchecked "
|
|
5827
|
+
+ "for its slot rules, never as admitted.",
|
|
5828
|
+
},
|
|
5829
|
+
{
|
|
5830
|
+
clauseRef: "C2",
|
|
5831
|
+
obligation: "A configuration violating each member of the vocabulary is refused, naming the field: an authored module-owned key by the "
|
|
5832
|
+
+ "three schemas and by the registry; a credential field naming an undeclared slot on every read, at every runtime's startup "
|
|
5833
|
+
+ "and by the CLI; an inherited environment naming a declared secret on read, at startup, at dispatch and by the CLI; the "
|
|
5834
|
+
+ "registry reads no variable a module does not declare; and every secret the framework hands an agent-spawning runtime is "
|
|
5835
|
+
+ "named so that the inherited-environment rule refuses it.",
|
|
5836
|
+
home: AssuranceEvidenceHome.OBSERVED,
|
|
5837
|
+
plane: AssuranceEvidenceExecutionPlane.VERIFICATION,
|
|
5838
|
+
homeRationale: "A refusal shows only when a configuration that violates it is driven, and a well-formed configuration looks the same "
|
|
5839
|
+
+ "whether the refusal exists or not. The driven tests load each violation through the real registry, the real CLI door "
|
|
5840
|
+
+ "and the agent dispatch; the forbidden set's completeness is measured over the framework-managed materials. Each carries "
|
|
5841
|
+
+ "an `assurance-evidence:` marker naming this clause.",
|
|
5842
|
+
},
|
|
5843
|
+
],
|
|
5844
|
+
minimumEvidence: "C1: per committed layer, the module-owned keys authored, the credential-naming fields read by field with any naming an "
|
|
5845
|
+
+ "undeclared slot or a declared secret, and the providers whose slot rules could not be checked. C2: the marked driven tests.",
|
|
5846
|
+
verificationSteps: [
|
|
5847
|
+
"Read the application's provider configurations, its own and each environment's, through the doors' own reader.",
|
|
5848
|
+
"For each provider, read its declared slots from its module, and name each provider whose module could not be read.",
|
|
5849
|
+
"Name every module-owned key authored, every credential field naming an undeclared slot, and every inherited environment naming a declared secret.",
|
|
5850
|
+
"Run the marked driven tests for C2.",
|
|
5851
|
+
],
|
|
5852
|
+
severityGuidance: "A committed configuration authoring a module-owned key, naming an undeclared slot or forwarding a declared secret is a MAJOR: "
|
|
5853
|
+
+ "it is refused at startup, so it is either a configuration that cannot run or one the refusal did not reach. A provider whose "
|
|
5854
|
+
+ "module could not be read, so its slot rules were not checked, is INSUFFICIENT EVIDENCE for that provider, not a finding; so "
|
|
5855
|
+
+ "is an inherited environment read while the provider declarations could not be, since a forwarded provider secret would then "
|
|
5856
|
+
+ "go unseen. A refusal a C2 test shows does not hold is a MAJOR.",
|
|
5857
|
+
limits: "The forbidden set is what the framework and the provider modules declare. A variable they do not declare, such as a "
|
|
5858
|
+
+ "developer's own shell credential that local development passes through to a coding agent, is not refused, and its name is "
|
|
5859
|
+
+ "not the framework's to know. The destination gate holds a provider to its module's host slot; a host whose name resolves "
|
|
5860
|
+
+ "to a loopback address is not caught by it (#2995). A configuration that never reaches a runtime or the CLI, such as one "
|
|
5861
|
+
+ "hydrated by a host the framework does not ship, is refused only by the registry's read path.",
|
|
5862
|
+
},
|
|
4302
5863
|
};
|
|
4303
5864
|
//# sourceMappingURL=assurance-evidence-requirement-bodies.definitions.js.map
|