@ztothez/design-engineering 2.0.0

package/knowledge-base/benchmarks/azure-optimizer/product-contract.yaml

@@ -0,0 +1,216 @@
+version: "1.1"
+id: azure-optimizer
+name: Azure Optimizer Infrastructure Analysis Dashboard
+status: benchmark
+purpose: >-
+  Enable an infrastructure operator to identify Azure resource risk and optimization opportunities,
+  inspect evidence and impact, run an analysis with truthful processing disclosure, choose a next
+  action, and export a result whose provenance remains inspectable.
+authority:
+  behavior: Current candidate behavior is evidence, while the normalized task model in SOURCE-EVIDENCE.md defines comparison intent.
+  visual: Candidate visuals are comparison inputs and never override task integrity, accessibility, responsive behavior, or evidence truth.
+  precedence:
+    - { path: SOURCE-EVIDENCE.md, role: Normalized owner-authorized product and comparison evidence, authority: primary }
+    - { path: acceptance-criteria.md, role: Task and evidence interpretation, authority: supporting }
+    - { path: anti-patterns.md, role: Azure-specific rejection patterns, authority: supporting }
+actors:
+  - id: infrastructure-operator
+    name: Infrastructure operator
+    goals:
+      - Identify the highest-priority infrastructure problem and its impact.
+      - Inspect evidence before accepting a recommendation.
+      - Distinguish live analysis from demo, cached, imported, or fallback results.
+    responsibilities:
+      - Verify environment, data origin, scope, and freshness.
+      - Validate remediation against organizational policy before implementation.
+      - Preserve evidence and provenance during export or handoff.
+outputs:
+  - id: estate-summary
+    name: Estate summary
+    sourceOfTruth: Current fixed comparison dataset or explicitly identified connected source.
+    requiredEvidence: [Environment, data origin, scope, freshness]
+  - id: prioritized-findings
+    name: Prioritized findings
+    sourceOfTruth: Structured resource observations and severity policy.
+    requiredEvidence: [Affected resource, observation, impact, evidence, severity, confidence]
+  - id: remediation-plan
+    name: Remediation plan
+    sourceOfTruth: Finding-specific recommendation and validation contract.
+    requiredEvidence: [Recommended action, expected effect, owner or destination, validation method]
+  - id: analysis-report
+    name: Analysis report
+    sourceOfTruth: Declared live, simulated, cached, or imported processing result.
+    requiredEvidence: [Processing origin, started and completed timestamps, limitations]
+  - id: provenance-export
+    name: Provenance-preserving export
+    sourceOfTruth: Current visible result and its evidence metadata.
+    requiredEvidence: [Environment, origin, freshness, findings, limitations]
+modes:
+  - id: overview
+    name: Overview
+    purpose: Understand estate context, priority, impact, and next actions.
+    input: Fixed comparison dataset or declared connected estate scope.
+    outputs: [estate-summary, prioritized-findings, remediation-plan]
+  - id: analysis
+    name: Analysis
+    purpose: Run the candidate analysis workflow and inspect processing origin and results.
+    input: Current declared environment and data source.
+    outputs: [analysis-report, prioritized-findings, remediation-plan]
+  - id: export
+    name: Export
+    purpose: Preserve the current result and provenance for review or handoff.
+    input: Current completed or imported result.
+    outputs: [provenance-export]
+stateMachines:
+  - id: analysis-run
+    initial: idle
+    states:
+      - { id: idle, category: idle, userVisible: true, requirement: "Data mode, scope, and available action are visible before execution." }
+      - { id: running, category: loading, userVisible: true, requirement: Processing path and progress are visible without implying completion. }
+      - { id: success, category: success, userVisible: true, requirement: "Result origin, freshness, evidence, findings, and actions are visible together." }
+      - { id: error, category: error, userVisible: true, requirement: "Failure cause, preserved context, and recovery action are visible." }
+      - { id: fallback, category: fallback, userVisible: true, requirement: Simulation or local fallback is persistently distinguished from live analysis. }
+      - { id: disconnected, category: terminal, userVisible: true, requirement: Disconnection is explicit and unsupported operational claims are absent. }
+    transitions:
+      - { from: idle, to: running, trigger: Start analysis }
+      - { from: running, to: success, trigger: Receive a complete declared result }
+      - { from: running, to: error, trigger: Backend or processing failure }
+      - { from: error, to: running, trigger: Retry preserved analysis }
+      - { from: error, to: fallback, trigger: Continue with explicitly labeled local or demo data }
+      - { from: running, to: disconnected, trigger: Connection becomes unavailable without fallback }
+      - { from: success, to: running, trigger: Run another analysis }
+metrics:
+  - id: resource-count
+    name: Tracked resources
+    definition: Number of resources included in the current declared estate scope.
+    sourceOfTruth: Current dataset resource inventory.
+    format: count
+    evidenceRequired: true
+    range: { min: 0, max: 1000000000 }
+  - id: potential-savings
+    name: Potential monthly savings
+    definition: Estimated recurring monthly reduction associated with evidenced recommendations.
+    sourceOfTruth: Sum of finding-level estimates under one declared currency, period, and baseline.
+    format: text
+    evidenceRequired: true
+  - id: issue-count
+    name: Issues found
+    definition: Number of current findings in the declared scope, grouped by labeled severity.
+    sourceOfTruth: Structured current findings collection.
+    format: count
+    evidenceRequired: true
+    range: { min: 0, max: 1000000000 }
+  - id: health-score
+    name: Infrastructure health
+    definition: Declared scoring function over current evidenced findings and telemetry.
+    sourceOfTruth: Versioned score formula with current inputs and limitations.
+    format: percentage
+    evidenceRequired: true
+    range: { min: 0, max: 100 }
+constraints:
+  outOfScope:
+    - Automatic production remediation without operator authorization.
+    - Claims about real Azure connectivity when only bundled or simulated data is available.
+    - General cloud optimization conclusions outside the declared evidence scope.
+  prohibitedClaims:
+    - Online, operational, connected, production, continuous-monitoring, or verified claims without current runtime evidence.
+    - Savings, health, issue, performance, or trend values without a declared source, scope, period, and freshness.
+    - Human approval or representative-user validation generated by an agent.
+  mockData:
+    allowedEnvironments: [development, test, demo]
+    mustBeLabeled: true
+    prohibitedInProduction: true
+    fallbackDisclosure: Simulation or local fallback must be disclosed before execution, during processing, in results, history, and exports.
+acceptanceCriteria:
+  - id: truthful-context
+    title: Environment and data provenance are explicit
+    severity: blocker
+    requirement: Users can identify environment, origin, connection state, scope, and freshness without consulting documentation.
+    evidence: [contract, runtime, screenshot, manual-review]
+    appliesToModes: [overview, analysis, export]
+  - id: priority-path
+    title: Priority, impact, evidence, and action form one path
+    severity: blocker
+    requirement: The highest-priority finding exposes affected scope, impact, supporting evidence, remediation, and validation method.
+    evidence: [contract, runtime, manual-review]
+    appliesToModes: [overview, analysis]
+  - id: honest-processing
+    title: Processing origin remains truthful
+    severity: blocker
+    requirement: Live, simulated, cached, imported, fallback, stale, and disconnected results cannot be mistaken for one another.
+    evidence: [contract, runtime, network, screenshot]
+    appliesToModes: [analysis, export]
+  - id: metric-integrity
+    title: Metrics support inspectable decisions
+    severity: blocker
+    requirement: Every visible metric has a definition, source, scope, period or timestamp, comparison baseline where relevant, and a supported decision.
+    evidence: [contract, runtime, manual-review]
+    appliesToModes: [overview, analysis]
+  - id: export-provenance
+    title: Export preserves visible provenance
+    severity: blocker
+    requirement: Exports retain environment, origin, freshness, findings, evidence references, and limitations from the current result.
+    evidence: [runtime, export]
+    appliesToModes: [export]
+  - id: responsive-integrity
+    title: Comparison task remains operable at required widths
+    severity: blocker
+    requirement: Context, metrics, findings, evidence, and actions remain readable without clipping, overlap, or incoherent competing scroll regions.
+    evidence: [runtime, screenshot, accessibility]
+    appliesToModes: [overview, analysis, export]
+verification:
+  journeyFile: journeys.json
+  viewports:
+    - { name: mobile-375, width: 375, height: 812 }
+    - { name: tablet-768, width: 768, height: 1024 }
+    - { name: desktop-1024, width: 1024, height: 768 }
+    - { name: wide-1440, width: 1440, height: 1000 }
+  bindings:
+    - { profile: overview-baseline, journey: overview-integrity, actor: infrastructure-operator, mode: overview, acceptanceCriteria: [truthful-context, priority-path, metric-integrity, responsive-integrity] }
+    - { profile: original-analysis, journey: run-original-analysis, actor: infrastructure-operator, mode: analysis, acceptanceCriteria: [honest-processing, priority-path] }
+    - { profile: lovable-analysis, journey: run-lovable-analysis, actor: infrastructure-operator, mode: analysis, acceptanceCriteria: [honest-processing, priority-path] }
+    - { profile: uiux-analysis, journey: run-uiux-analysis, actor: infrastructure-operator, mode: analysis, acceptanceCriteria: [honest-processing, priority-path] }
+    - { profile: ztothez-analysis, journey: run-ztothez-analysis, actor: infrastructure-operator, mode: analysis, acceptanceCriteria: [honest-processing, priority-path] }
+
+benchmark:
+  archetype: operational-dashboard
+  interface: browser
+  qualityDimensions:
+    - { id: truthful-disclosure, status: required, reason: "Data origin, environment, freshness, and processing mode drive trust." }
+    - { id: information-design, status: required, reason: "Operators must identify priority, impact, evidence, action, and validation." }
+    - { id: visual-system, status: required, reason: "Metrics and findings require a consistent operational hierarchy." }
+    - { id: accessibility, status: required, reason: "Analysis and findings must remain readable and operable without color alone." }
+    - { id: responsive-structure, status: required, reason: "The same decision path is verified across four viewport sizes." }
+    - { id: state-integrity, status: required, reason: "Live, fallback, stale, disconnected, and failed analysis states must not collapse." }
+    - { id: maintainability, status: required, reason: "Data processing, domain findings, state, and rendering require clear boundaries." }
+  tasks:
+    - id: prioritize-cloud-savings
+      primary: true
+      actor: infrastructure-operator
+      mode: analysis
+      intent: Run analysis, identify the highest-priority supported saving opportunity, inspect its evidence, and determine the next validation action.
+      start:
+        stateMachine: analysis-run
+        state: idle
+        observable: The operator can identify the current dataset and initiate analysis.
+      success:
+        stateMachine: analysis-run
+        state: success
+        observable: A prioritized finding exposes impact, source evidence, remediation, ownership, and validation guidance.
+        evidence: [contract, runtime, network]
+      recovery:
+        required: true
+        failure:
+          stateMachine: analysis-run
+          state: disconnected
+          observable: The interface identifies unavailable processing and does not present disconnected results as live.
+        observable: The operator can reconnect, retry, or explicitly choose labeled local fallback without losing context.
+      journey: { profile: ztothez-analysis, journey: run-ztothez-analysis }
+      browser: { route: /, narrowViewport: mobile-375 }
+  evidencePolicy:
+    missingEvidence: unverified
+    failedBehavior: failed
+    unsupportedCapability: limitation
+  comparison:
+    taskContractId: azure-optimizer-priority-analysis
+    crossContractRanking: false

package/knowledge-base/benchmarks/azure-optimizer/v2-human-review-methodology.yaml

@@ -0,0 +1,171 @@
+version: "1.1"
+id: azure-v2-anonymous-human-review
+title: Anonymous Azure interface quality and interaction comparison
+preparedAt: 2026-08-26T21:30:00+03:00
+preparedBy:
+  type: agent
+  name: ZtotheZ Design Engineering
+candidates:
+  - { id: anonymous-a, label: Candidate A }
+  - { id: anonymous-b, label: Candidate B }
+  - { id: anonymous-c, label: Candidate C }
+  - { id: anonymous-d, label: Candidate D }
+  - { id: anonymous-e, label: Candidate E }
+viewports:
+  - { id: mobile, width: 375, height: 812 }
+  - { id: tablet, width: 768, height: 1024 }
+  - { id: desktop, width: 1024, height: 768 }
+  - { id: wide, width: 1440, height: 1000 }
+states:
+  - { id: initial, label: Initial product state, required: true }
+  - { id: loading, label: Analysis in progress, required: true }
+  - { id: result, label: Completed result, required: true }
+  - { id: fallback, label: Backend failure with disclosed fallback, required: true }
+  - { id: disconnected, label: Disconnected state, required: true }
+  - { id: export, label: Exported result with provenance, required: true }
+stages:
+  - id: build
+    label: Production build
+    required: true
+    scope: candidate
+    evidenceKinds: [build-report]
+  - id: typecheck
+    label: Static type verification
+    required: true
+    scope: candidate
+    evidenceKinds: [typecheck-report]
+  - id: architecture-audit
+    label: Architecture and anti-slop audit
+    required: true
+    scope: candidate
+    evidenceKinds: [static-audit]
+  - id: browser-verification
+    label: Browser and responsive verification
+    required: true
+    scope: candidate
+    evidenceKinds: [runtime-report, screenshot]
+  - id: product-journeys
+    label: Product-specific journey verification
+    required: true
+    scope: candidate
+    evidenceKinds: [journey-report, export]
+  - id: human-review
+    label: Attributable human and representative-user review
+    required: true
+    scope: comparison
+    evidenceKinds: [human-review, user-session]
+tasks:
+  - id: identify-data-origin
+    prompt: Identify whether the displayed data is live, simulated, cached, imported, or unavailable, then locate environment and freshness.
+    successCriteria:
+      - Correctly identifies data origin and connection state.
+      - Locates freshness and environment without documentation.
+    measurements: [task-completion, time-on-task, navigation-errors, comprehension-accuracy, confidence]
+  - id: identify-priority
+    prompt: Find the highest-priority problem and explain its impact and supporting evidence.
+    successCriteria:
+      - Selects the intended highest-priority finding.
+      - Identifies impact and supporting evidence.
+    measurements: [task-completion, time-on-task, navigation-errors, comprehension-accuracy, confidence]
+  - id: identify-next-action
+    prompt: Identify the recommended next action, responsible owner, and validation method.
+    successCriteria:
+      - Locates the remediation action and owner.
+      - Identifies the stated validation method.
+    measurements: [task-completion, time-on-task, navigation-errors, recovery-attempts, confidence]
+  - id: recognize-fallback
+    prompt: Start analysis and determine whether live processing or a fallback produced the result.
+    successCriteria:
+      - Correctly identifies the processing path.
+      - Finds the result-level limitation or provenance statement.
+    measurements: [task-completion, time-on-task, navigation-errors, comprehension-accuracy, recovery-attempts, confidence]
+  - id: verify-export-provenance
+    prompt: Export the result and verify that data origin, freshness, environment, and limitations are retained.
+    successCriteria:
+      - Completes the export or correctly records that export is unavailable.
+      - Correctly identifies every retained or missing provenance field.
+    measurements: [task-completion, time-on-task, navigation-errors, comprehension-accuracy, recovery-attempts, confidence]
+rubric:
+  scale:
+    minimum: 0
+    maximum: 4
+    anchors:
+      "0": Absent, misleading, or unusable
+      "1": Present but materially deficient
+      "2": Functional and understandable
+      "3": Polished and consistently implemented
+      "4": Exceptional and supported by evidence
+  categories:
+    - id: truthful-disclosure
+      label: Truthful interface disclosure
+      weight: 30
+      criteria:
+        - id: disclosure-speed
+          prompt: Data origin, connection, environment, and freshness are identifiable without documentation.
+          measurement: comprehension
+          evidenceLevel: representative-user
+          scoreMinimum: 0
+          scoreMaximum: 4
+        - id: disclosure-consistency
+          prompt: Disclosure persists through action, result, history, fallback, error, and export states.
+          measurement: task-observation
+          evidenceLevel: human-expert
+          scoreMinimum: 0
+          scoreMaximum: 4
+    - id: information-design
+      label: Operational information design
+      weight: 35
+      criteria:
+        - id: priority-comprehension
+          prompt: Priority, impact, evidence, next action, and validation method form one understandable path.
+          measurement: comprehension
+          evidenceLevel: representative-user
+          scoreMinimum: 0
+          scoreMaximum: 4
+        - id: decision-density
+          prompt: Visible metrics and charts support identifiable decisions without decorative duplication.
+          measurement: rating
+          evidenceLevel: human-expert
+          scoreMinimum: 0
+          scoreMaximum: 4
+    - id: visual-polish
+      label: Visual hierarchy and polish
+      weight: 35
+      criteria:
+        - id: visual-rhythm
+          prompt: Spacing, alignment, typography, density, and emphasis appear intentional and coherent.
+          measurement: rating
+          evidenceLevel: human-expert
+          scoreMinimum: 0
+          scoreMaximum: 4
+        - id: responsive-composition
+          prompt: Composition remains balanced, legible, and task-focused at every required viewport.
+          measurement: task-observation
+          evidenceLevel: human-expert
+          scoreMinimum: 0
+          scoreMaximum: 4
+evidencePolicy:
+  anonymizeCandidates: true
+  counterbalanceOrder: true
+  retainArtifacts: true
+  separateEvidenceLevels: true
+  requireHumanAttribution: true
+  prohibitAgentHumanAttestation: true
+decision:
+  minimumCategoryScore: 3
+  requireAllStagesPass: true
+  prohibitContradictedClaims: true
+  requiredEvidenceLevels: [automated, ai-assisted-expert, human-expert, representative-user]
+  requiredStageCandidates: [anonymous-e]
+  reviewRequirements:
+    minimumHumanExpertSessions: 3
+    minimumRepresentativeUserSessions: 5
+    requireCompleteTaskMatrix: true
+    requireCompleteRatingMatrix: true
+    requireIdentityBlinding: true
+    minimumCounterbalancedOrders: 5
+  benchmark:
+    targetCandidate: anonymous-e
+    comparatorCandidates: [anonymous-b]
+    requiredCategoryIds: [truthful-disclosure, information-design, visual-polish]
+    requireTaskMetricNonRegression: true

package/knowledge-base/benchmarks/azure-optimizer/v2-journeys.json

@@ -0,0 +1,139 @@
+{
+  "version": "1.0",
+  "contract": "azure-optimizer",
+  "profiles": [
+    {
+      "id": "azure-v2-state-matrix",
+      "name": "Azure Optimizer V2 contract and state matrix",
+      "environment": [
+        "Azure Optimizer V2 production build is running",
+        "The fixed imported comparison fixture is unchanged",
+        "No Azure tenant connection is configured"
+      ],
+      "expectedNetwork": [
+        {
+          "id": "disclosed-fallback-failure",
+          "method": "POST",
+          "urlIncludes": "/api/analyze?scenario=fallback",
+          "status": 503,
+          "minOccurrences": 1,
+          "maxOccurrences": 1
+        }
+      ],
+      "journeys": [
+        {
+          "id": "demo-success",
+          "name": "demo-success",
+          "steps": [
+            { "action": "navigate", "value": "/?state=demo" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-state", "value": "normal" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-data-mode]", "name": "data-ztothez-design-data-mode", "value": "demo" },
+            { "action": "click", "selector": "[data-view=analysis]" },
+            { "action": "expectResponse", "selector": "#run-analysis", "urlIncludes": "/api/analyze?scenario=demo", "status": 200, "method": "POST" },
+            { "action": "waitFor", "selector": ".run-complete" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-stage", "value": "result" },
+            { "action": "expectText", "selector": "body", "value": "Analysis complete" }
+          ]
+        },
+        {
+          "id": "live-connected",
+          "name": "live-connected",
+          "steps": [
+            { "action": "navigate", "value": "/?state=live" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-data-mode]", "name": "data-ztothez-design-data-mode", "value": "live" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-connection]", "name": "data-ztothez-design-connection", "value": "connected" },
+            { "action": "click", "selector": "[data-view=analysis]" },
+            { "action": "expectResponse", "selector": "#run-analysis", "urlIncludes": "/api/analyze?scenario=live", "status": 200, "method": "POST" },
+            { "action": "waitFor", "selector": ".run-complete" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-result-origin]", "name": "data-ztothez-design-result-origin", "value": "live" }
+          ]
+        },
+        {
+          "id": "slow-analysis",
+          "name": "slow-analysis",
+          "steps": [
+            { "action": "navigate", "value": "/?state=slow" },
+            { "action": "click", "selector": "[data-view=analysis]" },
+            { "action": "click", "selector": "#run-analysis" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-stage", "value": "loading" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-state", "value": "slow" },
+            { "action": "waitFor", "selector": ".run-complete" },
+            { "action": "expectText", "selector": "body", "value": "Analysis complete" }
+          ]
+        },
+        {
+          "id": "backend-failure-fallback",
+          "name": "backend-failure-fallback",
+          "steps": [
+            { "action": "navigate", "value": "/?state=fallback" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-data-mode]", "name": "data-ztothez-design-data-mode", "value": "hybrid" },
+            { "action": "expectText", "selector": "[data-ztothez-design-result-origin]", "value": "fallback simulation" },
+            { "action": "click", "selector": "[data-view=analysis]" },
+            { "action": "expectResponse", "selector": "#run-analysis", "urlIncludes": "/api/analyze?scenario=fallback", "status": 503, "method": "POST" },
+            { "action": "waitFor", "selector": ".run-error" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-stage", "value": "error" },
+            { "action": "click", "selector": "#use-local-fallback" },
+            { "action": "waitFor", "selector": ".run-complete" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-result-origin]", "name": "data-ztothez-design-result-origin", "value": "simulated" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-state", "value": "fallback" }
+          ]
+        },
+        {
+          "id": "disconnected-recovery",
+          "name": "disconnected-recovery",
+          "steps": [
+            { "action": "navigate", "value": "/?state=disconnected" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-connection]", "name": "data-ztothez-design-connection", "value": "disconnected" },
+            { "action": "expectVisible", "selector": "[data-ztothez-design-recovery-action]" },
+            { "action": "click", "selector": ".trust-recovery" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-connection]", "name": "data-ztothez-design-connection", "value": "connected" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-state", "value": "normal" }
+          ]
+        },
+        {
+          "id": "partial-result",
+          "name": "partial-result",
+          "steps": [
+            { "action": "navigate", "value": "/?state=partial" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-state", "value": "partial" },
+            { "action": "expectText", "selector": "body", "value": "Partial result" },
+            { "action": "expectText", "selector": ".count-label", "value": "2 evidenced" }
+          ]
+        },
+        {
+          "id": "stale-result",
+          "name": "stale-result",
+          "steps": [
+            { "action": "navigate", "value": "/?state=stale" },
+            { "action": "expectAttribute", "selector": "[data-ztothez-design-freshness]", "name": "data-ztothez-design-freshness", "value": "stale" },
+            { "action": "expectText", "selector": "body", "value": "Stale result" }
+          ]
+        },
+        {
+          "id": "finding-detail-history",
+          "name": "finding-detail-history",
+          "steps": [
+            { "action": "navigate", "value": "/?state=demo" },
+            { "action": "click", "selector": "[data-finding-id=AZ-CONFIG-001]" },
+            { "action": "expectText", "selector": "#finding-detail-heading", "value": "backup policy" },
+            { "action": "click", "selector": "[data-view=history]" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-stage", "value": "history" },
+            { "action": "expectText", "selector": "body", "value": "No scheduled monitoring is implied" }
+          ]
+        },
+        {
+          "id": "export-provenance",
+          "name": "export-provenance",
+          "steps": [
+            { "action": "navigate", "value": "/?state=demo" },
+            { "action": "click", "selector": "[data-view=export]" },
+            { "action": "expectAttribute", "selector": "#azure-v2-app", "name": "data-ztothez-design-stage", "value": "export" },
+            { "action": "expectText", "selector": "body", "value": "processing origin" },
+            { "action": "expectDownload", "selector": "#download-provenance", "filenameIncludes": "provenance.json" },
+            { "action": "expectText", "selector": "body", "value": "Export created with provenance" }
+          ]
+        }
+      ]
+    }
+  ]
+}

package/knowledge-base/benchmarks/corpus/PROVENANCE.md

@@ -0,0 +1,21 @@
+# Corpus Provenance
+
+## Project-Owned Sources
+
+The AegisOPS and SceneStart benchmark contracts are derived from repositories and product evidence owned by ZtotheZ. Their source snapshots, authority boundaries, and transformation notes are recorded in their respective `SOURCE-EVIDENCE.md` and `MANIFEST.md` files.
+
+The compact architecture and anti-slop repositories under `cases/` were authored specifically for this project. They are synthetic behavioral fixtures, not extracted third-party application code. They are distributed under the repository license.
+
+## Licensed Standard Reference
+
+The visual-accessibility retrieval case refers to WCAG 2.2 concepts through independently authored query text and the independently authored ZtotheZ visual-accessibility module. It does not redistribute the WCAG specification.
+
+- Authoritative source: https://www.w3.org/TR/WCAG22/
+- License information: https://www.w3.org/copyright/document-license-2015/
+- Owner: World Wide Web Consortium
+- License record: W3C Document License 2015
+- Transformation: concepts are expressed as an original search query and expected local source path; no normative passage is copied into the case.
+
+## Inclusion Rules
+
+Every corpus case must reference one source declared in `corpus.yaml`. User-owned sources require local evidence. Licensed standards require an authoritative URL, license record, attribution, and a derivation statement. Reference archives and third-party design products are prohibited as corpus authorities.

package/knowledge-base/benchmarks/corpus/README.md

@@ -0,0 +1,30 @@
+# Corpus Benchmark
+
+This directory is the maintained evaluation corpus for ZtotheZ Design Engineering. It measures whether retrieval, architecture auditing, product-contract validation, and anti-slop rules produce the expected decisions on controlled positive and negative cases.
+
+## Files
+
+- `corpus.yaml` is the executable case manifest and score policy.
+- `corpus.schema.yaml` is the portable schema for corpus authors and external tooling.
+- `PROVENANCE.md` records ownership, licensed references, derivation, and redistribution boundaries.
+- `cases/` contains small project-owned repositories and one deliberately incomplete product contract.
+
+## Evaluation
+
+Run:
+
+```bash
+npm run evaluate-corpus
+```
+
+The command exits unsuccessfully when the overall threshold, a dimension threshold, or recommendation mean reciprocal rank is missed. A case passes only when the observed result matches its declared expectation. Negative cases pass when the system rejects the known defect or explicitly abstains.
+
+## Interpretation
+
+- Recommendation relevance measures whether the expected approved source appears within the declared maximum rank.
+- Abstention measures whether out-of-scope queries return `no-match` without archive fallback.
+- Architectural integrity measures clean acceptance and known architecture-rule detection.
+- Task completeness measures valid product contracts and rejection of broken cross-references.
+- Anti-slop rejection measures clean acceptance and deterministic rejection of mock production paths or nonfunctional controls.
+
+Passing this corpus proves behavior only for the included cases. It does not prove universal recommendation quality, product usability, legal clearance, or visual quality.

package/knowledge-base/benchmarks/corpus/cases/anti-slop-negative/package.json

@@ -0,0 +1,9 @@
+{
+  "name": "anti-slop-negative",
+  "private": true,
+  "scripts": {
+    "build": "echo build",
+    "test": "echo test",
+    "typecheck": "echo typecheck"
+  }
+}

package/knowledge-base/benchmarks/corpus/cases/anti-slop-negative/src/FakeDashboard.tsx

@@ -0,0 +1,10 @@
+export function FakeDashboard() {
+  const result = "fake progress result";
+  return (
+    <main>
+      <h1>Executive dashboard</h1>
+      <p>{result}</p>
+      <a href="#">Open full report</a>
+    </main>
+  );
+}

package/knowledge-base/benchmarks/corpus/cases/anti-slop-positive/package.json

@@ -0,0 +1,9 @@
+{
+  "name": "anti-slop-positive",
+  "private": true,
+  "scripts": {
+    "build": "echo build",
+    "test": "echo test",
+    "typecheck": "echo typecheck"
+  }
+}

package/knowledge-base/benchmarks/corpus/cases/anti-slop-positive/src/Progress.tsx

@@ -0,0 +1,15 @@
+type ProgressProps = {
+  completedIds: string[];
+  requiredIds: string[];
+};
+
+export function Progress({ completedIds, requiredIds }: ProgressProps) {
+  const completed = requiredIds.filter((id) => completedIds.includes(id)).length;
+  return (
+    <section aria-labelledby="progress-title">
+      <h2 id="progress-title">Release progress</h2>
+      <p>{completed} of {requiredIds.length} requirements complete</p>
+      <a href="/release/evidence">Inspect evidence</a>
+    </section>
+  );
+}

package/knowledge-base/benchmarks/corpus/cases/architecture-negative/package.json

@@ -0,0 +1,9 @@
+{
+  "name": "architecture-negative",
+  "private": true,
+  "scripts": {
+    "build": "echo build",
+    "test": "echo test",
+    "typecheck": "echo typecheck"
+  }
+}