diff --git a/.github/review-prompt.md b/.github/review-prompt.md
index ed511dc1..4f91e5ff 100644
--- a/.github/review-prompt.md
+++ b/.github/review-prompt.md
@@ -56,7 +56,8 @@ Use only the review contract in this file plus the generated scope and packet fi
 - Use `approved_with_notes` when only non-blocking follow-ups remain.
 - Use `changes_requested` when any blocking issue remains.
 - Fill the structured fields only. The renderer owns the markdown layout.
-- Keep `summary` to plain prose only. Do not include the PR title, a separate verdict line, markdown tables, file inventories, or custom section headings there.
+- Keep `summary` to plain prose only, ideally 2-4 sentences. Do not include the PR title, a separate verdict line, markdown tables, file inventories, or custom section headings there.
+- Keep `overallRationale` to 1 sentence.
 - Keep `what`, `why`, and `fix` concise plain text. Do not emit headings, tables, or fenced code blocks inside those fields.
 - Use `snippets` only when a short literal excerpt materially clarifies a finding. Keep each snippet under 20 lines, and do not include markdown fences in `code`.
 - Use `securityChecklist` for concise review rows about security-sensitive checks. Provide at least 1 row, and use 2-5 when possible. `status` = `pass` | `fail` | `na`.
diff --git a/.github/workflows/ai-review.yml b/.github/workflows/ai-review.yml
index 7ce457f0..bb3c2d0f 100644
--- a/.github/workflows/ai-review.yml
+++ b/.github/workflows/ai-review.yml
@@ -196,12 +196,12 @@ jobs:
     # GLM API environment for model routing
     env:
       ANTHROPIC_BASE_URL: https://api.z.ai/api/anthropic
-      REVIEW_MODEL: glm-5.1
+      REVIEW_MODEL: glm-5-turbo
       ANTHROPIC_AUTH_TOKEN: ${{ secrets.GLM_API_KEY }}
-      ANTHROPIC_MODEL: glm-5.1
-      ANTHROPIC_DEFAULT_OPUS_MODEL: glm-5.1
-      ANTHROPIC_DEFAULT_SONNET_MODEL: glm-5.1
-      ANTHROPIC_DEFAULT_HAIKU_MODEL: GLM-4.7-FlashX
+      ANTHROPIC_MODEL: glm-5-turbo
+      ANTHROPIC_DEFAULT_OPUS_MODEL: glm-5-turbo
+      ANTHROPIC_DEFAULT_SONNET_MODEL: glm-5-turbo
+      ANTHROPIC_DEFAULT_HAIKU_MODEL: glm-5-turbo
       DISABLE_BUG_COMMAND: '1'
       DISABLE_ERROR_REPORTING: '1'
       DISABLE_TELEMETRY: '1'
diff --git a/scripts/github/normalize-ai-review-output.mjs b/scripts/github/normalize-ai-review-output.mjs
index 9be73e21..06d2436c 100644
--- a/scripts/github/normalize-ai-review-output.mjs
+++ b/scripts/github/normalize-ai-review-output.mjs
@@ -14,6 +14,11 @@ const SEVERITY_HEADERS = {
   medium: '### 🟡 Medium',
   low: '### 🟢 Low',
 };
+const SEVERITY_SUMMARY_LABELS = {
+  high: '🔴 High',
+  medium: '🟡 Medium',
+  low: '🟢 Low',
+};
 
 const STATUS_LABELS = {
   pass: '✅',
@@ -41,6 +46,7 @@ const CODE_BLOCK_LANGUAGE_PATTERN = /^[A-Za-z0-9#+.-]{1,20}$/u;
 const MAX_FINDING_SNIPPETS = 2;
 const MAX_SNIPPET_LINES = 20;
 const MAX_SNIPPET_CHARACTERS = 1200;
+const TOP_FINDINGS_LIMIT = 3;
 
 function cleanText(value) {
   return typeof value === 'string' ? value.trim().replace(/\s+/g, ' ') : '';
@@ -224,35 +230,40 @@ function formatReviewContext(rendering) {
   const parts = [];
 
   if (rendering.mode) {
-    parts.push(`mode ${renderCode(rendering.mode)}`);
-    parts.push(REVIEW_MODE_DETAILS[rendering.mode]);
+    parts.push(renderCode(rendering.mode));
   }
 
-  const scopeSummary = formatScopeSummary(rendering);
-  if (scopeSummary) {
-    parts.push(`scope ${scopeSummary}`);
+  if (
+    typeof rendering.selectedFiles === 'number' &&
+    typeof rendering.reviewableFiles === 'number'
+  ) {
+    parts.push(`${rendering.selectedFiles}/${rendering.reviewableFiles} files`);
   }
 
-  const packetCoverage = formatPacketCoverage(rendering);
-  if (packetCoverage) {
-    parts.push(`packet ${packetCoverage}`);
+  if (
+    typeof rendering.selectedChanges === 'number' &&
+    typeof rendering.reviewableChanges === 'number'
+  ) {
+    parts.push(`${rendering.selectedChanges}/${rendering.reviewableChanges} lines`);
   }
 
-  const turnBudget = formatTurnBudget(rendering);
-  if (turnBudget) {
-    parts.push(`turn budget ${turnBudget}`);
+  if (
+    typeof rendering.packetIncludedFiles === 'number' &&
+    typeof rendering.packetTotalFiles === 'number'
+  ) {
+    parts.push(`packet ${rendering.packetIncludedFiles}/${rendering.packetTotalFiles}`);
   }
 
-  const timeBudget = formatTimeBudget(rendering);
-  if (timeBudget) {
-    parts.push(`workflow cap ${timeBudget}`);
+  const runtimeBudget = formatCombinedBudget(rendering) || formatTimeBudget(rendering) || formatTurnBudget(rendering);
+  if (runtimeBudget) {
+    parts.push(runtimeBudget);
   }
 
   if (parts.length === 0) {
     return null;
   }
 
-  return `> 🧭 Review context: ${parts.join('; ')}.`;
+  return `> 🧭 ${parts.join(' • ')}`;
 }
 
 function classifyFallbackReason(reason) {
@@ -612,8 +623,8 @@ export function normalizeStructuredOutput(raw) {
   return { ok: true, value };
 }
 
-function renderChecklistTable(title, labelHeader, labelKey, rows) {
-  const lines = ['', title, '', `| ${labelHeader} | Status | Notes |`, '|---|---|---|'];
+function renderChecklistTable(labelHeader, labelKey, rows) {
+  const lines = [`| ${labelHeader} | Status | Notes |`, '|---|---|---|'];
   for (const row of rows) {
     lines.push(
       `| ${renderInlineText(row[labelKey])} | ${STATUS_LABELS[row.status]} | ${renderInlineText(row.notes)} |`
@@ -622,9 +633,9 @@ function renderChecklistTable(title, labelHeader, labelKey, rows) {
   return lines;
 }
 
-function renderBulletSection(title, items) {
+function renderBulletSection(items) {
   if (items.length === 0) return [];
-  return ['', title, ...items.map((item) => `- ${renderInlineText(item)}`)];
+  return items.map((item) => `- ${renderInlineText(item)}`);
 }
 
 function renderFindingSnippets(snippets) {
@@ -646,46 +657,104 @@ function renderFindingSnippets(snippets) {
   return lines;
 }
 
+function renderDetailsBlock(summary, bodyLines) {
+  if (!bodyLines.length) {
+    return [];
+  }
+
+  return ['', '<details>', `<summary>${escapeMarkdown(summary)}</summary>`, '', ...bodyLines, '', '</details>'];
+}
+
+function renderFindingReference(finding) {
+  return finding.line ? `${finding.file}:${finding.line}` : finding.file;
+}
+
+function getOrderedFindings(findings) {
+  return SEVERITY_ORDER.flatMap((severity) => findings.filter((finding) => finding.severity === severity));
+}
+
+function renderTopFindings(findings) {
+  if (findings.length === 0) {
+    return ['No confirmed issues found after reviewing the diff and surrounding code.'];
+  }
+
+  const orderedFindings = getOrderedFindings(findings);
+  const lines = orderedFindings
+    .slice(0, TOP_FINDINGS_LIMIT)
+    .map(
+      (finding) =>
+        `- ${SEVERITY_SUMMARY_LABELS[finding.severity]} ${renderCode(renderFindingReference(finding))} — ${renderInlineText(finding.title)}`
+    );
+
+  if (orderedFindings.length > TOP_FINDINGS_LIMIT) {
+    const remaining = orderedFindings.length - TOP_FINDINGS_LIMIT;
+    lines.push(`- ${remaining} more finding${remaining === 1 ? '' : 's'} in the details below.`);
+  }
+
+  return lines;
+}
+
+function renderDetailedFindings(findings) {
+  if (findings.length === 0) {
+    return [];
+  }
+
+  const lines = [];
+  for (const severity of SEVERITY_ORDER) {
+    const scopedFindings = findings.filter((finding) => finding.severity === severity);
+    if (scopedFindings.length === 0) continue;
+
+    lines.push(`**${SEVERITY_SUMMARY_LABELS[severity]} (${scopedFindings.length})**`, '');
+    for (const finding of scopedFindings) {
+      lines.push(`- **${renderCode(renderFindingReference(finding))} — ${renderInlineText(finding.title)}**`);
+      lines.push(`  Problem: ${renderInlineText(finding.what)}`);
+      lines.push(`  Why it matters: ${renderInlineText(finding.why)}`);
+      lines.push(`  Suggested fix: ${renderInlineText(finding.fix)}`);
+      lines.push(...renderFindingSnippets(finding.snippets));
+      lines.push('');
+    }
+  }
+
+  if (lines[lines.length - 1] === '') {
+    lines.pop();
+  }
+
+  return lines;
+}
+
 export function renderStructuredReview(review, { model, rendering: renderOptions } = {}) {
   const rendering = mergeRenderingMetadata(review?.rendering, renderOptions);
-  const lines = ['### 📋 Summary', '', renderInlineText(review.summary), '', '### 🔍 Findings'];
+  const lines = [
+    '### Verdict',
+    '',
+    `**${ASSESSMENTS[review.overallAssessment]}** — ${renderInlineText(review.overallRationale)}`,
+    '',
+    renderInlineText(review.summary),
+  ];
   const reviewContext = formatReviewContext(rendering);
 
   if (reviewContext) {
-    lines.splice(4, 0, reviewContext, '');
+    lines.push('', reviewContext);
   }
 
-  if (review.findings.length === 0) {
-    lines.push('No confirmed issues found after reviewing the diff and surrounding code.');
-  } else {
-    for (const severity of SEVERITY_ORDER) {
-      const findings = review.findings.filter((finding) => finding.severity === severity);
-      if (findings.length === 0) continue;
-
-      lines.push('', SEVERITY_HEADERS[severity], '');
-      for (const finding of findings) {
-        const location = finding.line ? `${finding.file}:${finding.line}` : finding.file;
-        lines.push(`- **${renderCode(location)} — ${renderInlineText(finding.title)}**`);
-        lines.push(`  Problem: ${renderInlineText(finding.what)}`);
-        lines.push(`  Why it matters: ${renderInlineText(finding.why)}`);
-        lines.push(`  Suggested fix: ${renderInlineText(finding.fix)}`);
-        lines.push(...renderFindingSnippets(finding.snippets));
-        lines.push('');
-      }
-    }
-    if (lines[lines.length - 1] === '') lines.pop();
-  }
-
-  lines.push(...renderChecklistTable('### 🔒 Security Checklist', 'Check', 'check', review.securityChecklist));
-  lines.push(...renderChecklistTable('### 📊 CCS Compliance', 'Rule', 'rule', review.ccsCompliance));
-  lines.push(...renderBulletSection('### 💡 Informational', review.informational));
-  lines.push(...renderBulletSection("### ✅ What's Done Well", review.strengths));
+  lines.push('', '### Top Findings', '', ...renderTopFindings(review.findings));
+  lines.push(...renderDetailsBlock(`Full Findings (${review.findings.length})`, renderDetailedFindings(review.findings)));
+  lines.push(
+    ...renderDetailsBlock(
+      `Security Checklist (${review.securityChecklist.length})`,
+      renderChecklistTable('Check', 'check', review.securityChecklist)
+    )
+  );
+  lines.push(
+    ...renderDetailsBlock(
+      `CCS Compliance (${review.ccsCompliance.length})`,
+      renderChecklistTable('Rule', 'rule', review.ccsCompliance)
+    )
+  );
+  lines.push(...renderDetailsBlock(`Informational (${review.informational.length})`, renderBulletSection(review.informational)));
+  lines.push(...renderDetailsBlock(`What's Done Well (${review.strengths.length})`, renderBulletSection(review.strengths)));
 
   lines.push(
-    '',
-    '### 🎯 Overall Assessment',
-    '',
-    `**${ASSESSMENTS[review.overallAssessment]}** — ${renderInlineText(review.overallRationale)}`,
     '',
     `> 🤖 Reviewed by \`${model}\``
   );
diff --git a/scripts/github/run-ai-review-direct.mjs b/scripts/github/run-ai-review-direct.mjs
index a2f37959..c1604c8a 100644
--- a/scripts/github/run-ai-review-direct.mjs
+++ b/scripts/github/run-ai-review-direct.mjs
@@ -283,7 +283,7 @@ export async function writeDirectReviewFromEnv(env = process.env, fetchImpl = gl
       const responseJson = await postReviewRequest({
         apiUrl: cleanText(env.ANTHROPIC_BASE_URL),
         apiKey: cleanText(env.ANTHROPIC_AUTH_TOKEN),
-        model: cleanText(env.REVIEW_MODEL || env.ANTHROPIC_MODEL || 'glm-5.1'),
+        model: cleanText(env.REVIEW_MODEL || env.ANTHROPIC_MODEL || 'glm-5-turbo'),
         system,
         prompt: attemptPrompt,
         timeoutMs: attemptWindow.timeoutMs,
@@ -318,11 +318,11 @@ export async function writeDirectReviewFromEnv(env = process.env, fetchImpl = gl
 
   const markdown = finalValidation
     ? renderStructuredReview(finalValidation, {
-        model: cleanText(env.REVIEW_MODEL || 'glm-5.1'),
+        model: cleanText(env.REVIEW_MODEL || 'glm-5-turbo'),
         rendering,
       })
     : renderIncompleteReview({
-        model: cleanText(env.REVIEW_MODEL || 'glm-5.1'),
+        model: cleanText(env.REVIEW_MODEL || 'glm-5-turbo'),
         reason: lastReason,
         runUrl: cleanText(env.AI_REVIEW_RUN_URL || '#'),
         selectedFiles: coveredSelectedFiles,
diff --git a/tests/unit/scripts/github/normalize-ai-review-output.test.ts b/tests/unit/scripts/github/normalize-ai-review-output.test.ts
index afefb07e..c34d45e0 100644
--- a/tests/unit/scripts/github/normalize-ai-review-output.test.ts
+++ b/tests/unit/scripts/github/normalize-ai-review-output.test.ts
@@ -52,21 +52,22 @@ describe('normalize-ai-review-output', () => {
     );
 
     expect(validation.ok).toBe(true);
-    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5.1' });
+    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5-turbo' });
 
-    expect(markdown).toContain('### 📋 Summary');
-    expect(markdown).toContain('### 🔴 High');
+    expect(markdown).toContain('### Verdict');
+    expect(markdown).toContain('### Top Findings');
+    expect(markdown).toContain('- 🔴 High `src/cliproxy/accounts/query.ts:61` — Ambiguous account lookup drops valid matches');
+    expect(markdown).toContain('<summary>Full Findings (1)</summary>');
     expect(markdown).toContain('**`src/cliproxy/accounts/query.ts:61` — Ambiguous account lookup drops valid matches**');
-    expect(markdown).toContain('### 🔒 Security Checklist');
+    expect(markdown).toContain('<summary>Security Checklist (1)</summary>');
     expect(markdown).toContain('| Injection safety | ✅ | No user-controlled input reaches a shell, SQL, or HTML boundary in this diff. |');
-    expect(markdown).toContain('### 📊 CCS Compliance');
+    expect(markdown).toContain('<summary>CCS Compliance (1)</summary>');
     expect(markdown).toContain('| No emojis in CLI | N/A | This change affects GitHub PR comments only, not CLI stdout. |');
-    expect(markdown).toContain('### 💡 Informational');
-    expect(markdown).toContain('### ✅ What\'s Done Well');
-    expect(markdown).toContain('### 🎯 Overall Assessment');
+    expect(markdown).toContain('<summary>Informational (1)</summary>');
+    expect(markdown).toContain("<summary>What's Done Well (1)</summary>");
     expect(markdown).toContain('**❌ CHANGES REQUESTED**');
     expect(markdown).toContain('Why it matters: That breaks normal selection flows for users with multiple Codex sessions.');
-    expect(markdown).toContain('> 🤖 Reviewed by `glm-5.1`');
+    expect(markdown).toContain('> 🤖 Reviewed by `glm-5-turbo`');
   });
 
   test('renders mode-aware review context metadata without changing the structured review contract', () => {
@@ -97,7 +98,7 @@ describe('normalize-ai-review-output', () => {
 
     expect(validation.ok).toBe(true);
     const markdown = reviewOutput.renderStructuredReview(validation.value, {
-      model: 'glm-5.1',
+      model: 'glm-5-turbo',
       rendering: {
         mode: 'triage',
         selectedFiles: 8,
@@ -113,7 +114,7 @@ describe('normalize-ai-review-output', () => {
     });
 
     expect(markdown).toContain(
-      '> 🧭 Review context: mode `triage`; expanded packaged review with broader coverage; scope 8/34 reviewable files; 620/2140 reviewable changed lines; packet 6/8 selected files included in the final review packet; 2 selected files omitted for packet budget; turn budget 6 turns; workflow cap 5 minutes.'
+      '> 🧭 `triage` • 8/34 files • 620/2140 lines • packet 6/8 • 6 turns / 5 minutes'
     );
     expect(markdown).toContain('**⚠️ APPROVED WITH NOTES**');
   });
@@ -157,7 +158,7 @@ describe('normalize-ai-review-output', () => {
     );
 
     expect(validation.ok).toBe(true);
-    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5.1' });
+    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5-turbo' });
 
     expect(markdown).toContain('`buildReviewScope(files, mode)`');
     expect(markdown).toContain('`.github/workflows/ai-review.yml`');
@@ -200,7 +201,7 @@ describe('normalize-ai-review-output', () => {
     );
 
     expect(validation.ok).toBe(true);
-    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5.1' });
+    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5-turbo' });
 
     expect(markdown).toContain('Evidence: Current publish branch');
     expect(markdown).toContain('```bash');
@@ -242,7 +243,7 @@ describe('normalize-ai-review-output', () => {
     expect(validation.ok).toBe(true);
     expect(validation.value.findings[0].snippets[0].code).toBe('    if value:\n        print(value)');
 
-    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5.1' });
+    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5-turbo' });
     expect(markdown).toContain('    if value:');
     expect(markdown).toContain('        print(value)');
   });
@@ -300,7 +301,7 @@ describe('normalize-ai-review-output', () => {
 
       const result = reviewOutput.writeReviewFromEnv({
         AI_REVIEW_EXECUTION_FILE: executionFile,
-        AI_REVIEW_MODEL: 'glm-5.1',
+        AI_REVIEW_MODEL: 'glm-5-turbo',
         AI_REVIEW_MODE: 'triage',
         AI_REVIEW_SELECTED_FILES: '10',
         AI_REVIEW_REVIEWABLE_FILES: '46',
@@ -363,7 +364,7 @@ describe('normalize-ai-review-output', () => {
 
       const result = reviewOutput.writeReviewFromEnv({
         AI_REVIEW_EXECUTION_FILE: executionFile,
-        AI_REVIEW_MODEL: 'glm-5.1',
+        AI_REVIEW_MODEL: 'glm-5-turbo',
         AI_REVIEW_MODE: 'fast',
         AI_REVIEW_SELECTED_FILES: '6',
         AI_REVIEW_REVIEWABLE_FILES: '52',
@@ -410,7 +411,7 @@ describe('normalize-ai-review-output', () => {
 
       const result = reviewOutput.writeReviewFromEnv({
         AI_REVIEW_EXECUTION_FILE: executionFile,
-        AI_REVIEW_MODEL: 'glm-5.1',
+        AI_REVIEW_MODEL: 'glm-5-turbo',
         AI_REVIEW_OUTPUT_FILE: outputFile,
         AI_REVIEW_RUN_URL: 'https://github.com/kaitranntt/ccs/actions/runs/1',
         AI_REVIEW_STRUCTURED_OUTPUT: JSON.stringify({
@@ -507,15 +508,15 @@ describe('normalize-ai-review-output', () => {
     );
 
     expect(validation.ok).toBe(true);
-    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5.1' });
+    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5-turbo' });
 
-    expect(markdown).toContain('### 🔍 Findings');
+    expect(markdown).toContain('### Top Findings');
     expect(markdown).toContain('No confirmed issues found after reviewing the diff and surrounding code.');
-    expect(markdown).toContain('### 🔒 Security Checklist');
+    expect(markdown).toContain('<summary>Security Checklist (1)</summary>');
     expect(markdown).toContain(
       '| Injection safety | ✅ | No user-controlled data crosses a risky boundary in the reviewed diff. |'
     );
-    expect(markdown).toContain('### 📊 CCS Compliance');
+    expect(markdown).toContain('<summary>CCS Compliance (1)</summary>');
     expect(markdown).toContain('| Help/docs alignment | N/A | No CLI behavior changed, so there was nothing to update. |');
     expect(markdown).toContain('**✅ APPROVED** — No confirmed regressions or missing verification remain.');
   });
@@ -545,7 +546,7 @@ describe('normalize-ai-review-output', () => {
     );
 
     expect(validation.ok).toBe(true);
-    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5.1' });
+    const markdown = reviewOutput.renderStructuredReview(validation.value, { model: 'glm-5-turbo' });
 
     expect(markdown).toContain(
       '**`tests/unit/scripts/github/normalize-ai-review-output.test.ts` — Missing empty-state coverage**'
@@ -575,7 +576,7 @@ describe('normalize-ai-review-output', () => {
         overallAssessment: 'approved_with_notes',
         overallRationale: 'This is a formatting-only follow-up.',
       },
-      { model: 'glm-5.1' }
+      { model: 'glm-5-turbo' }
     );
 
     expect(markdown).toContain('**``src/weird`path.ts`` — Backtick-safe locations stay readable**');
diff --git a/tests/unit/scripts/github/run-ai-review-direct.test.ts b/tests/unit/scripts/github/run-ai-review-direct.test.ts
index 51331f3c..8131f3dd 100644
--- a/tests/unit/scripts/github/run-ai-review-direct.test.ts
+++ b/tests/unit/scripts/github/run-ai-review-direct.test.ts
@@ -60,7 +60,7 @@ describe('run-ai-review-direct', () => {
         {
           ANTHROPIC_BASE_URL: 'https://api.z.ai/api/anthropic',
           ANTHROPIC_AUTH_TOKEN: 'test-token',
-          REVIEW_MODEL: 'glm-5.1',
+          REVIEW_MODEL: 'glm-5-turbo',
           GITHUB_REPOSITORY: 'kaitranntt/ccs',
           AI_REVIEW_PROMPT: 'You are a reviewer.',
           AI_REVIEW_PACKET_FILE: packetFile,
@@ -110,7 +110,7 @@ describe('run-ai-review-direct', () => {
         {
           ANTHROPIC_BASE_URL: 'https://api.z.ai/api/anthropic',
           ANTHROPIC_AUTH_TOKEN: 'test-token',
-          REVIEW_MODEL: 'glm-5.1',
+          REVIEW_MODEL: 'glm-5-turbo',
           GITHUB_REPOSITORY: 'kaitranntt/ccs',
           AI_REVIEW_PROMPT: 'You are a reviewer.',
           AI_REVIEW_PACKET_FILE: packetFile,
@@ -147,11 +147,9 @@ describe('run-ai-review-direct', () => {
       );
 
       expect(result.usedFallback).toBe(false);
-      expect(fs.readFileSync(outputFile, 'utf8')).toContain('### 📋 Summary');
+      expect(fs.readFileSync(outputFile, 'utf8')).toContain('### Verdict');
       expect(fs.readFileSync(outputFile, 'utf8')).toContain('**✅ APPROVED**');
-      expect(fs.readFileSync(outputFile, 'utf8')).toContain(
-        'packet 1/1 selected files included in the final review packet'
-      );
+      expect(fs.readFileSync(outputFile, 'utf8')).toContain('packet 1/1');
       expect(JSON.parse(fs.readFileSync(logFile, 'utf8')).attempts).toHaveLength(1);
     });
   });
@@ -171,7 +169,7 @@ describe('run-ai-review-direct', () => {
         {
           ANTHROPIC_BASE_URL: 'https://api.z.ai/api/anthropic',
           ANTHROPIC_AUTH_TOKEN: 'test-token',
-          REVIEW_MODEL: 'glm-5.1',
+          REVIEW_MODEL: 'glm-5-turbo',
           GITHUB_REPOSITORY: 'kaitranntt/ccs',
           AI_REVIEW_PROMPT: 'You are a reviewer.',
           AI_REVIEW_PACKET_FILE: packetFile,
@@ -263,7 +261,7 @@ describe('run-ai-review-direct', () => {
         {
           ANTHROPIC_BASE_URL: 'https://api.z.ai/api/anthropic',
           ANTHROPIC_AUTH_TOKEN: 'test-token',
-          REVIEW_MODEL: 'glm-5.1',
+          REVIEW_MODEL: 'glm-5-turbo',
           GITHUB_REPOSITORY: 'kaitranntt/ccs',
           AI_REVIEW_PROMPT: 'You are a reviewer.',
           AI_REVIEW_PACKET_FILE: packetFile,
@@ -289,9 +287,7 @@ describe('run-ai-review-direct', () => {
 
       expect(result.usedFallback).toBe(false);
       expect(fs.readFileSync(outputFile, 'utf8')).toContain('**⚠️ APPROVED WITH NOTES**');
-      expect(fs.readFileSync(outputFile, 'utf8')).toContain(
-        'packet 2/3 selected files included in the final review packet; 1 selected file omitted for packet budget'
-      );
+      expect(fs.readFileSync(outputFile, 'utf8')).toContain('packet 2/3');
       expect(JSON.parse(fs.readFileSync(logFile, 'utf8')).attempts).toHaveLength(2);
     });
   });