@ryuenn3123/agentic-senior-core 3.0.19 → 3.0.21

package/scripts/ui-design-judge/design-execution-summary.mjs

@@ -44,6 +44,13 @@ function hasRepoEvidenceSummary(designIntentContent) {
   );
 }
 
+function hasStructuredInspectionEvidence(designIntentContent) {
+  return Boolean(
+    designIntentContent?.repoEvidence?.designEvidenceSummary?.structuredInspection
+    && typeof designIntentContent.repoEvidence.designEvidenceSummary.structuredInspection === 'object'
+  );
+}
+
 function summarizeDesignExecutionHandoff(designIntentContent) {
   const designExecutionHandoff = designIntentContent?.designExecutionHandoff
   && typeof designIntentContent.designExecutionHandoff === 'object'
@@ -72,6 +79,19 @@ function summarizeDesignExecutionHandoff(designIntentContent) {
     ? designExecutionHandoff.viewportMutationPlan
     : {};
 
+  function hasViewportMutationEntry(viewportKey) {
+    const viewportEntry = viewportMutationPlan?.[viewportKey];
+    return Boolean(
+      viewportEntry
+      && typeof viewportEntry === 'object'
+      && String(viewportEntry.primaryOperation || '').trim().length > 0
+      && Array.isArray(viewportEntry.requiredSurfaceActions)
+      && viewportEntry.requiredSurfaceActions.length > 0
+      && Array.isArray(viewportEntry.forbiddenPatterns)
+      && viewportEntry.forbiddenPatterns.length > 0
+    );
+  }
+
   const artifactChecks = [
     { name: 'surfacePlan', present: surfacePlan.length > 0 },
     { name: 'componentGraphNodes', present: componentGraphNodes.length > 1 },
@@ -82,7 +102,7 @@ function summarizeDesignExecutionHandoff(designIntentContent) {
     },
     {
       name: 'viewportMutationPlan',
-      present: ['mobile', 'tablet', 'desktop'].every((viewportKey) => String(viewportMutationPlan?.[viewportKey] || '').trim().length > 0),
+      present: ['mobile', 'tablet', 'desktop'].every((viewportKey) => hasViewportMutationEntry(viewportKey)),
     },
     { name: 'interactionStateMatrix', present: interactionStateMatrix.length > 0 },
     { name: 'taskFlowNarrative', present: taskFlowNarrative.length > 1 },
@@ -131,6 +151,7 @@ export function summarizeDesignExecutionPolicy(designIntentContent) {
     ? designExecutionPolicy.representationStrategy
     : null;
   const repoEvidenceAvailable = hasRepoEvidenceSummary(designIntentContent);
+  const structuredInspectionAvailable = hasStructuredInspectionEvidence(designIntentContent);
   const screenshotDependencyForbidden = designExecutionPolicy.forbidScreenshotDependency === true;
   const handoffFormatVersion = typeof designExecutionPolicy.handoffFormatVersion === 'string'
     ? designExecutionPolicy.handoffFormatVersion
@@ -170,6 +191,8 @@ export function summarizeDesignExecutionPolicy(designIntentContent) {
   }
   if (!repoEvidenceAvailable) {
     notes.push('repoEvidence.designEvidenceSummary is missing or unreadable.');
+  } else if (!structuredInspectionAvailable) {
+    notes.push('repoEvidence.designEvidenceSummary.structuredInspection is missing; class and inline-style evidence will stay lower-confidence.');
   }
   if (notes.length === 0) {
     notes.push('Structured design execution policy is present and ready for contract review.');
@@ -181,6 +204,7 @@ export function summarizeDesignExecutionPolicy(designIntentContent) {
     contractReady,
     screenshotDependencyForbidden,
     repoEvidenceAvailable,
+    structuredInspectionAvailable,
     handoffPresent: handoffSummary.present,
     handoffVersion: handoffSummary.version,
     handoffReady: handoffSummary.handoffReady,
@@ -216,8 +240,10 @@ export function summarizeReviewRubric(designIntentContent) {
   return {
     version: typeof reviewRubric.version === 'string' ? reviewRubric.version : null,
     dimensions,
+    genericityAutoFail: reviewRubric.genericityAutoFail === true,
     genericitySignals: normalizeStringArray(reviewRubric.genericitySignals),
     validBoldSignals: normalizeStringArray(reviewRubric.validBoldSignals),
+    forbiddenPatterns: normalizeStringArray(designIntentContent?.forbiddenPatterns),
     reportingRules: reviewRubric.reportingRules && typeof reviewRubric.reportingRules === 'object'
       ? {
         mustExplainGenericity: reviewRubric.reportingRules.mustExplainGenericity === true,

package/scripts/ui-design-judge/prompting.mjs

@@ -8,7 +8,7 @@ export function buildSystemPrompt() {
     'Compare the changed UI code against the provided design contract.',
     'Treat docs/design-intent.json as the machine-readable source of truth.',
     'Treat docs/DESIGN.md as explanatory context, not a generic style guide.',
-    'Treat designExecutionPolicy as the execution contract for how the UI should be planned, structured, and reviewed.',
+    'Treat designExecutionPolicy as the execution contract for how the UI must be planned, structured, and reviewed.',
     'Treat designExecutionHandoff as the explicit bridge between design intent and implementation decisions.',
     'Treat reviewRubric as the stable scoring frame for distinctiveness, contract fidelity, visual consistency, heuristic UX quality, and motion discipline.',
     'Use repoEvidence.designEvidenceSummary as implementation evidence when deciding whether the diff follows the intended system.',
@@ -17,15 +17,15 @@ export function buildSystemPrompt() {
     'Purposeful motion is allowed and can improve quality. Only flag motion when it drifts from the contract, ignores reduced-motion expectations, or adds avoidable performance/accessibility risk.',
     'Only flag drift when there is a clear mismatch with the contract, accessibility non-negotiables, or cross-viewport adaptation rules.',
     'Treat WCAG 2.2 AA failures as hard accessibility drift.',
-    'Treat APCA as advisory perceptual tuning only. Do not recommend blocking solely because APCA would prefer a stronger readability adjustment when WCAG hard requirements still pass.',
+    'Treat APCA as advisory perceptual tuning only. Do not set blocking solely because APCA indicates a stronger readability adjustment when WCAG hard requirements still pass.',
     'Check focus visibility, focus appearance, target size, keyboard access, accessible authentication, and status or dynamic state access when the diff touches those surfaces.',
-    'This audit always runs in advisory mode for this repository workflow.',
+    'Check design-intent.json for reviewRubric.genericityAutoFail. If true and forbiddenPatterns or genericitySignals are detected in the changed UI, this audit is no longer merely advisory for that finding: set blockingRecommended to true, mark the relevant rubric dimension as blocking, and require rebuilding the drifted surface instead of polishing it.',
     'Focus on color intent, typographic hierarchy, responsive re-layout, purposeful motion, component morphology across states, interaction behavior, and genericity drift.',
     'If you call something generic, explain the specific genericity signal or anti-pattern that caused that judgment.',
-    'Separate taste from failure. A bold design that follows the contract should not be penalized only because it is unusual.',
+    'Separate taste from failure. A bold design that follows the contract must not be penalized only because it is unusual.',
     'Return ONLY one JSON object on a single line prefixed with JSON_VERDICT:.',
     'Schema:',
-    '{"alignmentScore": number|null, "genericityAssessment": {"status": "distinctive|mixed|generic|unclear", "reason": string}, "tasteVsFailureSeparated": boolean, "rubricBreakdown": [{"dimension": string, "score": number|null, "verdict": "strong|acceptable|weak|unclear", "reason": string, "blocking": boolean}], "notes": string[], "findings": [{"area": string, "severity": "high|medium|low", "problem": string, "evidence": string, "recommendation": string, "blockingRecommended": boolean}]}',
+    '{"alignmentScore": number|null, "genericityAssessment": {"status": "distinctive|mixed|generic|unclear", "reason": string}, "tasteVsFailureSeparated": boolean, "rubricBreakdown": [{"dimension": string, "score": number|null, "verdict": "strong|acceptable|weak|unclear", "reason": string, "blocking": boolean}], "notes": string[], "findings": [{"area": string, "severity": "high|medium|low", "problem": string, "evidence": string, "requiredAction": string, "blockingRecommended": boolean}]}',
   ].join('\n');
 }
 

package/scripts/ui-design-judge/reporting.mjs

@@ -49,7 +49,7 @@ export function normalizeFindings(rawFindings) {
     severity: normalizeSeverity(rawFinding?.severity),
     problem: String(rawFinding?.problem || 'No problem description provided.'),
     evidence: String(rawFinding?.evidence || 'No evidence provided.'),
-    recommendation: String(rawFinding?.recommendation || 'No recommendation provided.'),
+    requiredAction: String(rawFinding?.requiredAction || 'No required action provided.'),
     blockingRecommended: rawFinding?.blockingRecommended === true,
   }));
 }
@@ -107,6 +107,7 @@ export function buildReport(partialReport) {
     mode: 'advisory',
     advisoryOnly: true,
     passed: true,
+    autoFailTriggered: false,
     skipped: false,
     skipReason: null,
     provider: 'none',
@@ -154,6 +155,7 @@ export function buildReport(partialReport) {
         statusChanged: false,
         namedGenericityRequired: false,
         matchedGenericitySignals: [],
+        matchedForbiddenPatterns: [],
         matchedValidBoldSignals: [],
         blockingFindingCount: 0,
         contractFidelityWeak: false,

package/scripts/ui-design-judge/rubric-calibration.mjs

@@ -24,7 +24,7 @@ function collectEvidenceTexts({
         finding?.area,
         finding?.problem,
         finding?.evidence,
-        finding?.recommendation,
+        finding?.requiredAction,
       ])
       : []),
     ...(Array.isArray(notes) ? notes : []),
@@ -93,6 +93,7 @@ export function calibrateGenericityAssessment({
   });
   const matchedGenericitySignals = collectMatchedSignals(reviewRubricSummary?.genericitySignals, evidenceTexts);
   const matchedValidBoldSignals = collectMatchedSignals(reviewRubricSummary?.validBoldSignals, evidenceTexts);
+  const matchedForbiddenPatterns = collectMatchedSignals(reviewRubricSummary?.forbiddenPatterns, evidenceTexts);
   const { blockingFindingCount, contractFidelityWeak, contractDriftDetected } = detectContractDrift(
     rubricBreakdown,
     findings,
@@ -106,17 +107,18 @@ export function calibrateGenericityAssessment({
     namedGenericityRequired
     && ['generic', 'mixed'].includes(providerStatus)
     && matchedGenericitySignals.length === 0
+    && matchedForbiddenPatterns.length === 0
   ) {
     calibratedStatus = 'unclear';
     calibrationNotes.push('Genericity claim was not backed by any named drift signal.');
   }
 
-  if (matchedGenericitySignals.length > 0 && matchedValidBoldSignals.length === 0) {
-    calibratedStatus = contractDriftDetected || matchedGenericitySignals.length >= 2
+  if ((matchedGenericitySignals.length > 0 || matchedForbiddenPatterns.length > 0) && matchedValidBoldSignals.length === 0) {
+    calibratedStatus = contractDriftDetected || matchedGenericitySignals.length + matchedForbiddenPatterns.length >= 2
       ? 'generic'
       : 'mixed';
     calibrationNotes.push('Named genericity drift signals dominate the review evidence.');
-  } else if (matchedValidBoldSignals.length > 0 && matchedGenericitySignals.length === 0) {
+  } else if (matchedValidBoldSignals.length > 0 && matchedGenericitySignals.length === 0 && matchedForbiddenPatterns.length === 0) {
     if (contractDriftDetected) {
       calibratedStatus = 'mixed';
       calibrationNotes.push('Authored signals are present, but contract drift prevents a distinctive verdict.');
@@ -127,7 +129,7 @@ export function calibrateGenericityAssessment({
       calibratedStatus = 'mixed';
       calibrationNotes.push('One valid bold signal was named, but evidence is not strong enough for a distinctive verdict.');
     }
-  } else if (matchedGenericitySignals.length > 0 && matchedValidBoldSignals.length > 0) {
+  } else if ((matchedGenericitySignals.length > 0 || matchedForbiddenPatterns.length > 0) && matchedValidBoldSignals.length > 0) {
     calibratedStatus = contractDriftDetected ? 'mixed' : 'mixed';
     calibrationNotes.push('The evidence contains both generic drift and legitimate authored moves.');
   } else if (providerStatus === 'distinctive' && contractDriftDetected) {
@@ -151,6 +153,7 @@ export function calibrateGenericityAssessment({
     statusChanged: calibratedStatus !== providerStatus,
     namedGenericityRequired,
     matchedGenericitySignals,
+    matchedForbiddenPatterns,
     matchedValidBoldSignals,
     blockingFindingCount,
     contractFidelityWeak,

package/scripts/ui-design-judge/rubric-goldset.json

@@ -77,7 +77,7 @@
           "severity": "high",
           "problem": "Mobile layout no longer follows the contract.",
           "evidence": "The primary action now drops below supporting proof.",
-          "recommendation": "Restore the intended task order.",
+          "requiredAction": "Restore the intended task order.",
           "blockingRecommended": true
         }
       ],
@@ -115,7 +115,7 @@
           "severity": "high",
           "problem": "The page uses balanced card grid without priority shift.",
           "evidence": "Every block carries equal weight and the CTA no longer leads.",
-          "recommendation": "Rebuild the page around one dominant task surface.",
+          "requiredAction": "Rebuild the page around one dominant task surface.",
           "blockingRecommended": true
         }
       ],

package/scripts/ui-design-judge.mjs

@@ -4,10 +4,13 @@
 /**
  * ui-design-judge.mjs
  *
- * Advisory-first UI design contract judge.
+ * Advisory-default UI design contract judge.
  *
  * Repo-internal workflow audit; no user-facing runtime modes.
- * Runs only in advisory mode for this repository workflow.
+ * Stays advisory when no provider is configured or no auto-fail signal is found.
+ * reviewRubric.genericityAutoFail escalates named drift signals and forbidden
+ * patterns into blocking findings and blocking required actions when provider
+ * evidence is available.
  *
  * Validation anchors for repo governance:
  * - Do not reward generic SaaS defaults or popular template patterns.
@@ -17,6 +20,7 @@
  * - designExecutionPolicy
  * - designExecutionHandoff
  * - handoffReady
+ * - structuredInspectionAvailable
  */
 
 import { collectChangedFiles, collectPullRequestDiff, isUiRelevantFilePath } from './ui-design-judge/git-input.mjs';
@@ -33,6 +37,56 @@ import {
 } from './ui-design-judge/reporting.mjs';
 import { loadDesignGuide, loadDesignIntent, summarizeDesignExecutionPolicy, summarizeReviewRubric } from './ui-design-judge/design-execution-summary.mjs';
 
+function applyGenericityAutoFail({
+  reviewRubricSummary,
+  calibration,
+  findings,
+  notes,
+}) {
+  const autoFailEnabled = reviewRubricSummary?.genericityAutoFail === true;
+  const namedGenericityDetected = Array.isArray(calibration?.matchedGenericitySignals)
+    && calibration.matchedGenericitySignals.length > 0;
+  const forbiddenPatternDetected = Array.isArray(calibration?.matchedForbiddenPatterns)
+    && calibration.matchedForbiddenPatterns.length > 0;
+  const shouldAutoFail = autoFailEnabled
+    && calibration?.calibratedStatus === 'generic'
+    && (namedGenericityDetected || forbiddenPatternDetected);
+
+  if (!shouldAutoFail) {
+    return {
+      findings,
+      notes,
+      autoFailTriggered: false,
+    };
+  }
+
+  const normalizedFindings = Array.isArray(findings) ? findings.map((finding) => ({ ...finding })) : [];
+  const updatedFindings = normalizedFindings.length > 0
+    ? normalizedFindings.map((finding) => ({
+      ...finding,
+      blockingRecommended: true,
+    }))
+    : [
+      {
+        area: 'design-contract',
+        severity: 'high',
+        problem: 'The UI matches named genericity drift signals that the contract marks as auto-fail.',
+        evidence: `Matched signals: ${[...(calibration.matchedGenericitySignals || []), ...(calibration.matchedForbiddenPatterns || [])].join(', ')}`,
+        requiredAction: 'Rebuild the affected UI surfaces from the contract and remove the named generic patterns instead of polishing them.',
+        blockingRecommended: true,
+      },
+    ];
+
+  return {
+    findings: updatedFindings,
+    notes: [
+      ...(Array.isArray(notes) ? notes : []),
+      'reviewRubric.genericityAutoFail triggered because named genericity drift signals or forbidden patterns were detected.',
+    ],
+    autoFailTriggered: true,
+  };
+}
+
 async function main() {
   const changedFiles = collectChangedFiles();
   const changedUiFiles = changedFiles.filter(isUiRelevantFilePath);
@@ -100,6 +154,7 @@ async function main() {
         driftCount: 0,
         blockingCandidateCount: 0,
         designExecutionSignalCount: designExecutionSummary.enabledCapabilities.length,
+        structuredInspectionAvailable: designExecutionSummary.structuredInspectionAvailable,
         genericityStatus: calibration.calibratedStatus,
       },
       designExecution: designExecutionSummary,
@@ -154,6 +209,7 @@ async function main() {
         driftCount: 0,
         blockingCandidateCount: 0,
         designExecutionSignalCount: designExecutionSummary.enabledCapabilities.length,
+        structuredInspectionAvailable: designExecutionSummary.structuredInspectionAvailable,
         genericityStatus: calibration.calibratedStatus,
       },
       designExecution: designExecutionSummary,
@@ -188,7 +244,6 @@ async function main() {
   const tasteVsFailureSeparated = typeof verdict?.tasteVsFailureSeparated === 'boolean'
     ? verdict.tasteVsFailureSeparated
     : null;
-  const blockingCandidateCount = findings.filter((finding) => finding.blockingRecommended || finding.severity === 'high').length;
   const alignmentScore = typeof verdict?.alignmentScore === 'number' ? verdict.alignmentScore : null;
   const notes = Array.isArray(verdict?.notes)
     ? verdict.notes.map((note) => String(note))
@@ -202,18 +257,31 @@ async function main() {
     notes,
     tasteVsFailureSeparated,
   });
+  const autoFailResolution = applyGenericityAutoFail({
+    reviewRubricSummary,
+    calibration,
+    findings,
+    notes,
+  });
+  const resolvedFindings = autoFailResolution.findings;
+  const resolvedNotes = autoFailResolution.notes;
+  const blockingCandidateCount = resolvedFindings.filter((finding) => finding.blockingRecommended || finding.severity === 'high').length;
 
   emitMachineReadableReport(buildReport({
     provider: selectedProvider.providerName,
     contractPresent: true,
-    passed: true,
+    mode: autoFailResolution.autoFailTriggered ? 'blocking-recommended' : 'advisory',
+    advisoryOnly: !autoFailResolution.autoFailTriggered,
+    passed: !autoFailResolution.autoFailTriggered,
+    autoFailTriggered: autoFailResolution.autoFailTriggered,
     malformedVerdict: malformed,
     summary: {
       changedUiFileCount: changedUiFiles.length,
       alignmentScore,
-      driftCount: findings.length,
+      driftCount: resolvedFindings.length,
       blockingCandidateCount,
       designExecutionSignalCount: designExecutionSummary.enabledCapabilities.length,
+      structuredInspectionAvailable: designExecutionSummary.structuredInspectionAvailable,
       genericityStatus: calibration.calibratedStatus,
     },
     designExecution: designExecutionSummary,
@@ -229,10 +297,10 @@ async function main() {
       skipped: false,
       skipReason: null,
     },
-    findings,
+    findings: resolvedFindings,
     notes: malformed
       ? ['LLM response was malformed. Advisory mode kept the audit non-blocking.', ...designExecutionSummary.notes]
-      : [...notes, ...calibration.notes, ...designExecutionSummary.notes],
+      : [...resolvedNotes, ...calibration.notes, ...designExecutionSummary.notes],
   }));
 }