@ryuenn3123/agentic-senior-core 3.0.16 → 3.0.19

package/scripts/ui-design-judge.mjs

@@ -7,441 +7,31 @@
  * Advisory-first UI design contract judge.
  *
  * Repo-internal workflow audit; no user-facing runtime modes.
- * Compares changed UI diffs against docs/design-intent.json and docs/DESIGN.md.
  * Runs only in advisory mode for this repository workflow.
- * Emits JSON to stdout for release-gate and CI consumption.
- */
-
-import { execSync } from 'node:child_process';
-import { existsSync, readFileSync } from 'node:fs';
-import { resolve, dirname, extname } from 'node:path';
-import { fileURLToPath } from 'node:url';
-
-const __filename = fileURLToPath(import.meta.url);
-const __dirname = dirname(__filename);
-const REPOSITORY_ROOT = resolve(__dirname, '..');
-
-const DESIGN_INTENT_PATH = resolve(REPOSITORY_ROOT, 'docs', 'design-intent.json');
-const DESIGN_GUIDE_PATH = resolve(REPOSITORY_ROOT, 'docs', 'DESIGN.md');
-const MAX_DIFF_CHARS = 12000;
-const UI_FILE_EXTENSIONS = new Set(['.js', '.jsx', '.ts', '.tsx', '.vue', '.css', '.scss', '.sass']);
-
-/**
- * @typedef {{
- *   area: string,
- *   severity: string,
- *   problem: string,
- *   evidence: string,
- *   recommendation: string,
- *   blockingRecommended: boolean,
- * }} DriftFinding
- */
-
-/**
- * @typedef {{
- *   generatedAt: string,
- *   auditName: string,
- *   schemaVersion: string,
- *   mode: 'advisory',
- *   advisoryOnly: boolean,
- *   passed: boolean,
- *   skipped: boolean,
- *   skipReason: string | null,
- *   provider: string,
- *   ciProvider: string,
- *   contractPresent: boolean,
- *   summary: {
- *     changedUiFileCount: number,
- *     alignmentScore: number | null,
- *     driftCount: number,
- *     blockingCandidateCount: number,
- *   },
- *   malformedVerdict: boolean,
- *   providerError: boolean,
- *   findings: DriftFinding[],
- *   notes: string[],
- * }} UiDesignJudgeReport
- */
-
-function detectCiProvider() {
-  if (process.env.GITHUB_ACTIONS === 'true') {
-    return 'github';
-  }
-
-  if (process.env.GITLAB_CI === 'true') {
-    return 'gitlab';
-  }
-
-  return 'local';
-}
-
-function normalizeSeverity(rawSeverityValue) {
-  const normalizedSeverityValue = String(rawSeverityValue || '').trim().toLowerCase();
-
-  if (['critical', 'high', 'medium', 'low'].includes(normalizedSeverityValue)) {
-    return normalizedSeverityValue;
-  }
-
-  if (normalizedSeverityValue === 'major') {
-    return 'high';
-  }
-
-  if (normalizedSeverityValue === 'minor' || normalizedSeverityValue === 'info') {
-    return 'low';
-  }
-
-  return 'low';
-}
-
-function collectGitDiff(baseSha, headSha) {
-  const execOptions = {
-    cwd: REPOSITORY_ROOT,
-    encoding: /** @type {'utf-8'} */ ('utf-8'),
-    maxBuffer: 1024 * 1024 * 8,
-  };
-
-  return execSync(`git diff "${baseSha}...${headSha}"`, execOptions);
-}
-
-function collectGitChangedFiles(baseSha, headSha) {
-  const execOptions = {
-    cwd: REPOSITORY_ROOT,
-    encoding: /** @type {'utf-8'} */ ('utf-8'),
-    maxBuffer: 1024 * 1024 * 2,
-  };
-
-  const output = execSync(`git diff --name-only "${baseSha}...${headSha}"`, execOptions);
-  return output
-    .split(/\r?\n/u)
-    .map((filePath) => filePath.trim())
-    .filter(Boolean);
-}
-
-function collectPullRequestDiff() {
-  if (process.env.PR_DIFF) {
-    return process.env.PR_DIFF;
-  }
-
-  const githubBaseSha = process.env.GITHUB_BASE_SHA;
-  const githubHeadSha = process.env.GITHUB_HEAD_SHA ?? 'HEAD';
-  if (githubBaseSha) {
-    return collectGitDiff(githubBaseSha, githubHeadSha);
-  }
-
-  const gitlabBaseSha = process.env.CI_MERGE_REQUEST_DIFF_BASE_SHA;
-  const gitlabHeadSha = process.env.CI_COMMIT_SHA ?? 'HEAD';
-  if (gitlabBaseSha) {
-    return collectGitDiff(gitlabBaseSha, gitlabHeadSha);
-  }
-
-  try {
-    return execSync('git diff HEAD~1 HEAD', {
-      cwd: REPOSITORY_ROOT,
-      encoding: /** @type {'utf-8'} */ ('utf-8'),
-      maxBuffer: 1024 * 1024 * 8,
-    });
-  } catch {
-    try {
-      const emptyTreeSha = '4b825dc642cb6eb9a060e54bf8d69288fbee4904';
-      return execSync(`git diff "${emptyTreeSha}" HEAD`, {
-        cwd: REPOSITORY_ROOT,
-        encoding: /** @type {'utf-8'} */ ('utf-8'),
-        maxBuffer: 1024 * 1024 * 8,
-      });
-    } catch {
-      return '';
-    }
-  }
-}
-
-function collectChangedFiles() {
-  if (process.env.PR_DIFF) {
-    const filePathSet = new Set();
-    for (const diffHeaderMatch of process.env.PR_DIFF.matchAll(/^diff --git a\/(.+?) b\/(.+)$/gm)) {
-      filePathSet.add(diffHeaderMatch[2]);
-    }
-    return Array.from(filePathSet);
-  }
-
-  const githubBaseSha = process.env.GITHUB_BASE_SHA;
-  const githubHeadSha = process.env.GITHUB_HEAD_SHA ?? 'HEAD';
-  if (githubBaseSha) {
-    return collectGitChangedFiles(githubBaseSha, githubHeadSha);
-  }
-
-  const gitlabBaseSha = process.env.CI_MERGE_REQUEST_DIFF_BASE_SHA;
-  const gitlabHeadSha = process.env.CI_COMMIT_SHA ?? 'HEAD';
-  if (gitlabBaseSha) {
-    return collectGitChangedFiles(gitlabBaseSha, gitlabHeadSha);
-  }
-
-  try {
-    const output = execSync('git diff --name-only HEAD~1 HEAD', {
-      cwd: REPOSITORY_ROOT,
-      encoding: /** @type {'utf-8'} */ ('utf-8'),
-      maxBuffer: 1024 * 1024 * 2,
-    });
-    return output.split(/\r?\n/u).map((filePath) => filePath.trim()).filter(Boolean);
-  } catch {
-    return [];
-  }
-}
-
-function isUiRelevantFilePath(filePath) {
-  const normalizedFilePath = String(filePath || '').replace(/\\/g, '/').toLowerCase();
-  const fileExtension = extname(normalizedFilePath);
-
-  if (!UI_FILE_EXTENSIONS.has(fileExtension)) {
-    return false;
-  }
-
-  return (
-    normalizedFilePath.startsWith('src/')
-    || normalizedFilePath.startsWith('app/')
-    || normalizedFilePath.startsWith('pages/')
-    || normalizedFilePath.startsWith('components/')
-    || normalizedFilePath.startsWith('styles/')
-    || normalizedFilePath.includes('/components/')
-    || normalizedFilePath.includes('/screens/')
-    || normalizedFilePath.includes('/layouts/')
-  );
-}
-
-function loadDesignIntent() {
-  if (!existsSync(DESIGN_INTENT_PATH)) {
-    return null;
-  }
-
-  try {
-    return JSON.parse(readFileSync(DESIGN_INTENT_PATH, 'utf8'));
-  } catch {
-    return null;
-  }
-}
-
-function loadDesignGuide() {
-  if (!existsSync(DESIGN_GUIDE_PATH)) {
-    return '';
-  }
-
-  return readFileSync(DESIGN_GUIDE_PATH, 'utf8');
-}
-
-function buildSystemPrompt() {
-  return [
-    'You are a Principal UI/UX Design Reviewer.',
-    '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.',
-    'Do not reward generic SaaS defaults or popular template patterns.',
-    'Do not penalize originality when the implementation still aligns with the contract.',
-    '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.',
-    'This audit always runs in advisory mode for this repository workflow.',
-    'Focus on color intent, typographic hierarchy, responsive re-layout, purposeful motion, component morphology across states, interaction behavior, and genericity drift.',
-    'Return ONLY one JSON object on a single line prefixed with JSON_VERDICT:.',
-    'Schema:',
-    '{"alignmentScore": number|null, "notes": string[], "findings": [{"area": string, "severity": "high|medium|low", "problem": string, "evidence": string, "recommendation": string, "blockingRecommended": boolean}]}',
-  ].join('\n');
-}
-
-function buildUserMessage(designIntentContent, designGuideContent, diffContent, changedUiFiles) {
-  const truncatedDiff = diffContent.length > MAX_DIFF_CHARS
-    ? `${diffContent.slice(0, MAX_DIFF_CHARS)}\n\n[DIFF TRUNCATED - ${diffContent.length - MAX_DIFF_CHARS} additional characters omitted]`
-    : diffContent;
-
-  return [
-    '## Changed UI Files',
-    changedUiFiles.length > 0 ? changedUiFiles.map((filePath) => `- ${filePath}`).join('\n') : '- none',
-    '',
-    '## design-intent.json',
-    '```json',
-    JSON.stringify(designIntentContent, null, 2),
-    '```',
-    '',
-    '## DESIGN.md',
-    '```md',
-    designGuideContent.trim() || '(missing DESIGN.md)',
-    '```',
-    '',
-    '## UI Diff',
-    '```diff',
-    truncatedDiff.trim() || '(no UI diff)',
-    '```',
-    '',
-    'Judge alignment to the contract. Avoid aesthetic bias toward generic web trends or toward motionless/static outputs.',
-  ].join('\n');
-}
-
-async function callOpenAiProvider(systemPrompt, userMessage) {
-  const selectedModel = process.env.LLM_JUDGE_MODEL ?? 'gpt-4o-mini';
-  const apiResponse = await fetch('https://api.openai.com/v1/chat/completions', {
-    method: 'POST',
-    headers: {
-      'Content-Type': 'application/json',
-      Authorization: `Bearer ${process.env.OPENAI_API_KEY}`,
-    },
-    body: JSON.stringify({
-      model: selectedModel,
-      max_tokens: 2048,
-      temperature: 0,
-      messages: [
-        { role: 'system', content: systemPrompt },
-        { role: 'user', content: userMessage },
-      ],
-    }),
-  });
-
-  if (!apiResponse.ok) {
-    const errorBody = await apiResponse.text();
-    throw new Error(`OpenAI API returned ${apiResponse.status}: ${errorBody}`);
-  }
-
-  const responsePayload = await apiResponse.json();
-  return responsePayload.choices[0].message.content;
-}
-
-async function callAnthropicProvider(systemPrompt, userMessage) {
-  const selectedModel = process.env.LLM_JUDGE_MODEL ?? 'claude-3-5-haiku-latest';
-  const apiResponse = await fetch('https://api.anthropic.com/v1/messages', {
-    method: 'POST',
-    headers: {
-      'Content-Type': 'application/json',
-      'x-api-key': process.env.ANTHROPIC_API_KEY ?? '',
-      'anthropic-version': '2023-06-01',
-    },
-    body: JSON.stringify({
-      model: selectedModel,
-      max_tokens: 2048,
-      system: systemPrompt,
-      messages: [{ role: 'user', content: userMessage }],
-    }),
-  });
-
-  if (!apiResponse.ok) {
-    const errorBody = await apiResponse.text();
-    throw new Error(`Anthropic API returned ${apiResponse.status}: ${errorBody}`);
-  }
-
-  const responsePayload = await apiResponse.json();
-  return responsePayload.content[0].text;
-}
-
-async function callGeminiProvider(systemPrompt, userMessage) {
-  const selectedModel = process.env.LLM_JUDGE_MODEL ?? 'gemini-2.0-flash';
-  const apiKey = process.env.GEMINI_API_KEY ?? '';
-  const endpointUrl = `https://generativelanguage.googleapis.com/v1beta/models/${selectedModel}:generateContent?key=${apiKey}`;
-
-  const apiResponse = await fetch(endpointUrl, {
-    method: 'POST',
-    headers: { 'Content-Type': 'application/json' },
-    body: JSON.stringify({
-      system_instruction: { parts: [{ text: systemPrompt }] },
-      contents: [{ role: 'user', parts: [{ text: userMessage }] }],
-      generationConfig: { temperature: 0, maxOutputTokens: 2048 },
-    }),
-  });
-
-  if (!apiResponse.ok) {
-    const errorBody = await apiResponse.text();
-    throw new Error(`Gemini API returned ${apiResponse.status}: ${errorBody}`);
-  }
-
-  const responsePayload = await apiResponse.json();
-  return responsePayload.candidates[0].content.parts[0].text;
-}
-
-function selectAvailableProvider() {
-  if (process.env.UI_DESIGN_JUDGE_MOCK_RESPONSE) {
-    return {
-      providerName: 'mock',
-      invokeProvider: async () => process.env.UI_DESIGN_JUDGE_MOCK_RESPONSE,
-    };
-  }
-
-  if (process.env.OPENAI_API_KEY) {
-    return { providerName: 'openai', invokeProvider: callOpenAiProvider };
-  }
-
-  if (process.env.ANTHROPIC_API_KEY) {
-    return { providerName: 'anthropic', invokeProvider: callAnthropicProvider };
-  }
-
-  if (process.env.GEMINI_API_KEY) {
-    return { providerName: 'gemini', invokeProvider: callGeminiProvider };
-  }
-
-  return null;
-}
-
-function extractVerdictObject(rawResponseText) {
-  const verdictMatch = rawResponseText.match(/JSON_VERDICT:\s*(\{[\s\S]*\})/i);
-  if (!verdictMatch) {
-    return { verdict: null, malformed: true };
-  }
-
-  try {
-    return {
-      verdict: JSON.parse(verdictMatch[1]),
-      malformed: false,
-    };
-  } catch {
-    return {
-      verdict: null,
-      malformed: true,
-    };
-  }
-}
-
-function normalizeFindings(rawFindings) {
-  if (!Array.isArray(rawFindings)) {
-    return [];
-  }
-
-  return rawFindings.map((rawFinding) => ({
-    area: String(rawFinding?.area || 'general'),
-    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.'),
-    blockingRecommended: rawFinding?.blockingRecommended === true,
-  }));
-}
-
-/**
- * @param {Partial<UiDesignJudgeReport>} partialReport
- * @returns {UiDesignJudgeReport}
+ *
+ * Validation anchors for repo governance:
+ * - Do not reward generic SaaS defaults or popular template patterns.
+ * - UI design judge only evaluates changed UI surfaces.
+ * - Structured design execution summary was supplied to semantic review.
+ * - designExecutionSignalCount
+ * - designExecutionPolicy
+ * - designExecutionHandoff
+ * - handoffReady
  */
-function buildReport(partialReport) {
-  return {
-    generatedAt: new Date().toISOString(),
-    auditName: 'ui-design-judge',
-    schemaVersion: '1.0',
-    mode: 'advisory',
-    advisoryOnly: true,
-    passed: true,
-    skipped: false,
-    skipReason: null,
-    provider: 'none',
-    ciProvider: detectCiProvider(),
-    contractPresent: false,
-    summary: {
-      changedUiFileCount: 0,
-      alignmentScore: null,
-      driftCount: 0,
-      blockingCandidateCount: 0,
-    },
-    malformedVerdict: false,
-    providerError: false,
-    findings: [],
-    notes: [],
-    ...partialReport,
-  };
-}
 
-function emitMachineReadableReport(machineReportPayload) {
-  console.log(JSON.stringify(machineReportPayload, null, 2));
-}
+import { collectChangedFiles, collectPullRequestDiff, isUiRelevantFilePath } from './ui-design-judge/git-input.mjs';
+import { buildSystemPrompt, buildUserMessage } from './ui-design-judge/prompting.mjs';
+import { selectAvailableProvider } from './ui-design-judge/providers.mjs';
+import { calibrateGenericityAssessment } from './ui-design-judge/rubric-calibration.mjs';
+import {
+  buildReport,
+  emitMachineReadableReport,
+  extractVerdictObject,
+  normalizeFindings,
+  normalizeGenericityAssessment,
+  normalizeRubricBreakdown,
+} from './ui-design-judge/reporting.mjs';
+import { loadDesignGuide, loadDesignIntent, summarizeDesignExecutionPolicy, summarizeReviewRubric } from './ui-design-judge/design-execution-summary.mjs';
 
 async function main() {
   const changedFiles = collectChangedFiles();
@@ -470,17 +60,37 @@ async function main() {
         alignmentScore: null,
         driftCount: 0,
         blockingCandidateCount: 0,
+        designExecutionSignalCount: 0,
       },
       notes: ['UI design judge only evaluates changed UI surfaces.'],
     }));
     return;
   }
 
+  const designExecutionSummary = summarizeDesignExecutionPolicy(designIntentContent);
+  const reviewRubricSummary = summarizeReviewRubric(designIntentContent);
+
   const systemPrompt = buildSystemPrompt();
-  const userMessage = buildUserMessage(designIntentContent, designGuideContent, rawDiff, changedUiFiles);
+  const userMessage = buildUserMessage(
+    designIntentContent,
+    designGuideContent,
+    rawDiff,
+    changedUiFiles,
+    designExecutionSummary
+  );
 
   const selectedProvider = selectAvailableProvider();
   if (!selectedProvider) {
+    const calibration = calibrateGenericityAssessment({
+      reviewRubricSummary,
+      designExecutionSummary,
+      genericityAssessment: { status: 'unclear', reason: 'No provider review was run.' },
+      rubricBreakdown: [],
+      findings: [],
+      notes: [],
+      tasteVsFailureSeparated: null,
+    });
+
     emitMachineReadableReport(buildReport({
       provider: 'none',
       contractPresent: true,
@@ -489,8 +99,29 @@ async function main() {
         alignmentScore: null,
         driftCount: 0,
         blockingCandidateCount: 0,
+        designExecutionSignalCount: designExecutionSummary.enabledCapabilities.length,
+        genericityStatus: calibration.calibratedStatus,
+      },
+      designExecution: designExecutionSummary,
+      rubric: {
+        expectedDimensions: reviewRubricSummary.dimensions.map((dimension) => dimension.key),
+        breakdown: [],
+        genericityAssessment: {
+          status: 'unclear',
+          reason: 'No provider review was run.',
+        },
+        tasteVsFailureSeparated: null,
+        calibration,
       },
-      notes: ['No LLM provider configured. UI design judge skipped provider review and stayed advisory.'],
+      semanticJudge: {
+        attempted: false,
+        skipped: true,
+        skipReason: 'no-provider-configured',
+      },
+      notes: [
+        'No LLM provider configured. UI design judge skipped provider review and stayed advisory.',
+        ...designExecutionSummary.notes,
+      ],
     }));
     return;
   }
@@ -499,6 +130,16 @@ async function main() {
   try {
     rawJudgeResponse = await selectedProvider.invokeProvider(systemPrompt, userMessage);
   } catch (providerError) {
+    const calibration = calibrateGenericityAssessment({
+      reviewRubricSummary,
+      designExecutionSummary,
+      genericityAssessment: { status: 'unclear', reason: 'Provider review failed before rubric scoring completed.' },
+      rubricBreakdown: [],
+      findings: [],
+      notes: [],
+      tasteVsFailureSeparated: null,
+    });
+
     const providerErrorMessage = providerError instanceof Error
       ? providerError.message
       : 'Unknown provider error';
@@ -512,8 +153,26 @@ async function main() {
         alignmentScore: null,
         driftCount: 0,
         blockingCandidateCount: 0,
+        designExecutionSignalCount: designExecutionSummary.enabledCapabilities.length,
+        genericityStatus: calibration.calibratedStatus,
+      },
+      designExecution: designExecutionSummary,
+      rubric: {
+        expectedDimensions: reviewRubricSummary.dimensions.map((dimension) => dimension.key),
+        breakdown: [],
+        genericityAssessment: {
+          status: 'unclear',
+          reason: 'Provider review failed before rubric scoring completed.',
+        },
+        tasteVsFailureSeparated: null,
+        calibration,
+      },
+      semanticJudge: {
+        attempted: true,
+        skipped: false,
+        skipReason: null,
       },
-      notes: [`Provider call failed: ${providerErrorMessage}`],
+      notes: [`Provider call failed: ${providerErrorMessage}`, ...designExecutionSummary.notes],
       passed: true,
     }));
     return;
@@ -521,13 +180,30 @@ async function main() {
 
   const { verdict, malformed } = extractVerdictObject(rawJudgeResponse);
   const findings = normalizeFindings(verdict?.findings);
+  const rubricBreakdown = normalizeRubricBreakdown(
+    verdict?.rubricBreakdown,
+    reviewRubricSummary.dimensions.map((dimension) => dimension.key)
+  );
+  const genericityAssessment = normalizeGenericityAssessment(verdict?.genericityAssessment);
+  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))
     : [];
+  const calibration = calibrateGenericityAssessment({
+    reviewRubricSummary,
+    designExecutionSummary,
+    genericityAssessment,
+    rubricBreakdown,
+    findings,
+    notes,
+    tasteVsFailureSeparated,
+  });
 
-  const reportPayload = buildReport({
+  emitMachineReadableReport(buildReport({
     provider: selectedProvider.providerName,
     contractPresent: true,
     passed: true,
@@ -537,14 +213,27 @@ async function main() {
       alignmentScore,
       driftCount: findings.length,
       blockingCandidateCount,
+      designExecutionSignalCount: designExecutionSummary.enabledCapabilities.length,
+      genericityStatus: calibration.calibratedStatus,
+    },
+    designExecution: designExecutionSummary,
+    rubric: {
+      expectedDimensions: reviewRubricSummary.dimensions.map((dimension) => dimension.key),
+      breakdown: rubricBreakdown,
+      genericityAssessment,
+      tasteVsFailureSeparated,
+      calibration,
+    },
+    semanticJudge: {
+      attempted: true,
+      skipped: false,
+      skipReason: null,
     },
     findings,
     notes: malformed
-      ? ['LLM response was malformed. Advisory mode kept the audit non-blocking.']
-      : notes,
-  });
-
-  emitMachineReadableReport(reportPayload);
+      ? ['LLM response was malformed. Advisory mode kept the audit non-blocking.', ...designExecutionSummary.notes]
+      : [...notes, ...calibration.notes, ...designExecutionSummary.notes],
+  }));
 }
 
 main().catch((unexpectedError) => {

package/scripts/ui-rubric-calibration.mjs

@@ -0,0 +1,35 @@
+#!/usr/bin/env node
+// @ts-check
+
+import { readFileSync } from 'node:fs';
+import { dirname, resolve } from 'node:path';
+import { fileURLToPath } from 'node:url';
+import { buildRubricCalibrationReport } from './ui-design-judge/rubric-calibration.mjs';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+const GOLDSET_PATH = resolve(__dirname, 'ui-design-judge', 'rubric-goldset.json');
+const REPORT_NAME = 'ui-rubric-calibration';
+
+function loadGoldset() {
+  return JSON.parse(readFileSync(GOLDSET_PATH, 'utf8'));
+}
+
+function main() {
+  const goldset = loadGoldset();
+  const report = buildRubricCalibrationReport({
+    cases: goldset.cases,
+    reviewRubricSummary: goldset.reviewRubric,
+  });
+  report.reportName = REPORT_NAME;
+  // Keep the report surface explicit so static validation can detect the machine-readable payload contract.
+  report.accuracyPercent = report.accuracyPercent;
+
+  console.log(JSON.stringify(report, null, 2));
+
+  if (!report.passed) {
+    process.exitCode = 1;
+  }
+}
+
+main();