fix(linkedin-studio): ship placeholder voice profile, gitignore real, sentinel detection

Wave 2 / Step 5 of the remediation plan (coupled criticals: voice-leak +
placeholder-detection).

Voice profile (the adopter-default leak):
- Ship a PII-free placeholder at authentic-voice-samples.md carrying a
  <!-- VOICE_PLACEHOLDER --> sentinel + neutral default voice principles.
- Migrate the author's real profile to gitignored authentic-voice-samples.local.md
  (already matched by *.local.md; added an explicit, commented .gitignore entry so
  the intent is unmissable). NO git-history rewrite — the historical file is
  attributed authorship, not a secret (per the plan threat model).
- Add authentic-voice-samples.template.md — a clean fill-in template for adopters.
- personalization-score.mjs: detect the sentinel (deterministic) instead of the
  unreliable `[Your Name]` heuristic, so the placeholder scores 0 voice points and
  a populated profile (sentinel removed) earns the 25.
- Both voice writers replace-not-append on the placeholder: setup.md (merge ->
  replace-if-placeholder) and onboarding.md (append -> replace-if-placeholder), so
  populating removes the sentinel; updated setup.md's stale heuristic table.

Operator decisions (deviations from plan-literal, approved this session):
- KEEP the plugin.json author name. The plan said scrub author -> neutral/org, but
  that contradicts its own LICENSE reasoning (intentional MIT attribution) and all
  5 sibling plugins keep author = the author; scrubbing only this one would create
  inconsistency for zero security gain (the name is public-by-design). The voice
  placeholder fully fixes the adopter-inheritance bug.
- Scrub the stale "January 2026 360Brew" brand from the plugin.json description and
  the "360brew" keyword (locked decision: no publishable model name/date). This is
  a Wave-1 propagation miss surfaced here because plugin.json was in Step 5's
  touch-scope.

Flagged for follow-up (NOT done here — out of Session 2 scope):
- The lint's stat-consistency grep (scripts/test-runner.sh) scans references/,
  commands/, skills/, hooks/prompts/, CLAUDE.md, README.md — but NOT
  .claude-plugin/plugin.json, which is why the 360Brew brand slipped Wave 1.
  Needs a Session-1-scoped lint extension to add plugin.json to the scan set.
- Readers (user-prompt-context.mjs, voice-guardian.md, state-update-reminder.md)
  read the tracked .md (placeholder), per the plan. The operator's real voice now
  lives in the gitignored .local.md, which nothing reads. To use it, readers + the
  voice score should prefer .local.md (matching the user-profile.local.md
  precedent). Deferred as a coherence follow-up for operator review.

Test-first: hooks/scripts/__tests__/personalization-score.test.mjs (red on the
placeholder scoring 25 under the old heuristic, green after the sentinel fix). Hook
suite 62/62, structural lint 0 failed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

plugins/linkedin-studio/.claude-plugin/plugin.json

@@ -1,11 +1,11 @@
 {
   "name": "linkedin-studio",
   "version": "3.1.0",
-  "description": "LinkedIn Studio — full-spectrum LinkedIn content engine: feed posts, carousels, video scripts, and long-form newsletter editions, with the January 2026 360Brew algorithm baked in. v3.1.0 (Endring 9) adds a cold adversarial review package to `/linkedin:newsletter` — Step 6.5 + the standalone `/linkedin:headless-review` command run three new headless archetypes (content-reviewer, language-reviewer, fact-reviewer) plus the persona reviewer with NO drafting-session context — a `/linkedin:pivot` command that re-opens cleared gates after a late change, and per-artifact personas (one or more readers configurable per edition). v3.0.0 renamed the plugin (was `linkedin-thought-leadership`): slug, agent namespace, and state-file path are `linkedin-studio`; the `/linkedin:*` commands are unchanged.",
+  "description": "LinkedIn Studio — full-spectrum LinkedIn content engine: feed posts, carousels, video scripts, and long-form newsletter editions, with the 2026 relevance-ranking model baked in. v3.1.0 (Endring 9) adds a cold adversarial review package to `/linkedin:newsletter` — Step 6.5 + the standalone `/linkedin:headless-review` command run three new headless archetypes (content-reviewer, language-reviewer, fact-reviewer) plus the persona reviewer with NO drafting-session context — a `/linkedin:pivot` command that re-opens cleared gates after a late change, and per-artifact personas (one or more readers configurable per edition). v3.0.0 renamed the plugin (was `linkedin-thought-leadership`): slug, agent namespace, and state-file path are `linkedin-studio`; the `/linkedin:*` commands are unchanged.",
   "author": {
     "name": "Kjell Tore Guttormsen"
   },
   "license": "MIT",
   "repository": "https://git.fromaitochitta.com/open/ktg-plugin-marketplace",
-  "keywords": ["linkedin", "content-creation", "newsletter", "analytics", "360brew"]
+  "keywords": ["linkedin", "content-creation", "newsletter", "analytics", "relevance-ranking"]
 }

plugins/linkedin-studio/.gitignore

@@ -5,6 +5,10 @@
 
 # Local configuration
 *.local.md
+# Real voice profile is personal data — adopters keep theirs local; the tracked
+# authentic-voice-samples.md ships as a sentinel placeholder. (Already matched by
+# *.local.md above; listed explicitly so the intent is unmissable.)
+assets/voice-samples/authentic-voice-samples.local.md
 
 # Session state (personal activity, auto-initialized from template)
 REMEMBER.md

plugins/linkedin-studio/assets/voice-samples/authentic-voice-samples.md

@@ -1,227 +1,70 @@
-# Authentic Voice Samples - Kjell Tore Guttormsen
+<!-- VOICE_PLACEHOLDER -->
+<!--
+  This is the SHIPPED PLACEHOLDER voice profile — neutral defaults, not anyone's
+  personal voice. The VOICE_PLACEHOLDER sentinel above keeps your voice
+  personalization score at 0 until you replace this file with your own profile.
 
-These guidelines help Claude understand and replicate Kjell Tore's natural writing style for LinkedIn content.
+  To personalize: run `/linkedin:setup` (Voice samples) or `/linkedin:onboarding`.
+  Those workflows overwrite this file with a profile built from your own samples.
+  Prefer to start from a clean form? Copy `authentic-voice-samples.template.md`.
 
-## Voice Profile Summary
+  If you want to keep your real profile out of version control, save it as
+  `authentic-voice-samples.local.md` (gitignored) instead of editing this file.
+-->
 
-Kjell Tore does not have traditional writing samples to share. Instead, his voice is defined by the following characteristics which Claude should internalize and apply consistently.
+# Authentic Voice Samples — Placeholder (neutral defaults)
 
----
+These are neutral, widely-applicable defaults so the plugin produces reasonable
+content before you personalize. Replace them with your own voice via
+`/linkedin:setup`. Until you do, the voice category scores 0.
 
 ## Core Voice Characteristics
 
-### 1. Solution-Oriented Mindset
+1. **Solution-oriented** — frame problems with a path forward, not just complaints.
-- Every problem is presented as an opportunity
+2. **Factually grounded** — base claims on evidence; acknowledge uncertainty openly.
-- Never complains without offering a path forward
+3. **Non-judgmental** — explain without criticizing people, companies, or decisions.
-- Focuses on "what can be done" rather than "what went wrong"
+4. **Curious and open** — treat "I don't know" as a starting point, not a weakness.
-- Sees challenges as interesting puzzles to solve
+5. **Story-led** — open with a concrete example before the abstract point.
+6. **Actionable** — end with something the reader can do, or a clear takeaway.
 
-### 2. Factual Grounding
+## Do's
-- Statements are based on facts, not assumptions
-- If uncertain, acknowledges uncertainty openly
-- Prefers data and evidence over opinions
-- Avoids speculation presented as fact
 
-### 3. Non-Judgmental Tone
+- ✅ Open with a story or concrete example before the concept.
-- Observes and explains without criticizing others
+- ✅ Use clear, accessible language even for technical topics.
-- Builds up, never tears down
+- ✅ Explain jargon on first use — assume intelligence, not prior knowledge.
-- Avoids negative commentary about people, companies, or decisions
+- ✅ Show rather than tell.
-- When discussing alternatives, frames as "different approaches" not "better/worse"
+- ✅ End with a specific, actionable takeaway.
+- ✅ Keep standard posts concise (≈800–1500 characters).
 
-### 4. Curiosity and Openness
+## Don'ts
-- Genuinely interested in learning new things
-- Open to new ideas and approaches
-- Asks questions to understand, not to challenge
-- Embraces "I don't know" as a starting point for exploration
 
-### 5. Storytelling Approach
+- ❌ Corporate buzzwords ("game-changer", "leverage", "synergy", "disrupt").
-- Uses narrative techniques to make points memorable
+- ❌ Criticizing people, companies, or decisions.
-- Varies storytelling patterns based on content:
+- ❌ Claims without evidence.
-  - Hero's journey (transformation stories)
+- ❌ More than 1–2 emojis per post.
-  - Problem-solution (practical content)
+- ❌ Generic motivational filler or preachy lecturing.
-  - Before-after (showing change/improvement)
-  - Discovery narrative (learning something new)
-  - Day-in-the-life (practical application)
-- Shows rather than tells
-
-### 6. Actionable Conclusions
-- Ends with something the reader can do
-- The more actionable, the better
-- If no clear action, provides a clear summary/takeaway
-- Never ends on a vague note
-
----
-
-## Cross-Sample Analysis
-
-### Do's (Things that sound like Kjell Tore)
-
-- ✅ Start with stories or concrete examples before explaining concepts
-- ✅ Use clear, accessible language even for technical topics
-- ✅ Explain technical concepts thoroughly - assume intelligence, not knowledge
-- ✅ Show rather than tell - demonstrate with examples
-- ✅ End with actionable takeaways - what can the reader do NOW?
-- ✅ Vary storytelling techniques based on the content
-- ✅ Be genuinely helpful and supportive
-- ✅ Acknowledge complexity before simplifying
-- ✅ Use transitions like "What I've learned is..." to share insights
-- ✅ Frame discoveries as shared learning, not lecturing
-- ✅ Keep posts concise - short to medium length (800-1500 characters)
-
-### Don'ts (Things Kjell Tore would NEVER say)
-
-- ❌ Don't use buzzwords: "game-changer", "leverage", "synergy", "disrupt", "revolutionize"
-- ❌ Don't criticize people, companies, or decisions
-- ❌ Don't use self-deprecating humor
-- ❌ Don't make assumptions without facts
-- ❌ Don't write overly long posts (stay under 1500 characters for posts)
-- ❌ Don't use more than 1-2 emojis per post
-- ❌ Don't discuss politics, religion, or personal matters
-- ❌ Don't use em dashes (—) - use hyphens or alternatives instead
-- ❌ Don't start with "Let's dive deep into..."
-- ❌ Don't use excessive exclamation marks!!!
-- ❌ Don't use generic motivational phrases
-- ❌ Don't be preachy or lecture the reader
-- ❌ Don't use "we" when you mean "I" (be direct about personal experience)
-
----
 
 ## Signature Phrases
 
-Use these naturally when appropriate - don't force them:
+_(Add the phrases you naturally use once you personalize this profile.)_
-
-- "Let me show you..."
-- "What I've learned is..."
-- "Here is the secret to..."
-
-These phrases signal a transition to insight or demonstration. Use them to introduce key points or revelations.
-
----
 
 ## Vocabulary Preferences
 
-### Technical Terms - How to Handle
+_(List the terms you always explain, and the words/phrases you avoid.)_
-
-- **RAG (Retrieval-Augmented Generation):** Always explain on first use
-- **MCP (Model Context Protocol):** Explain what it enables, not just the acronym
-- **Copilot Studio:** Can assume some familiarity with Microsoft ecosystem
-- **Skills (Claude):** Explain as "reusable instruction sets" or similar
-- **Low-code:** Generally understood, but clarify scope if needed
-
-**Principle:** Assume intelligence, not knowledge. Explain jargon without being condescending.
-
-### Words/Phrases to AVOID
-
-- "Game-changer"
-- "Revolutionary"
-- "Disruption" / "Disruptive"
-- "Leverage" (as a verb)
-- "Synergy"
-- "Deep dive" / "Let's dive deep"
-- "Unpack" (as in "let me unpack this")
-- "At the end of the day"
-- "It is what it is"
-- "Touch base"
-- "Circle back"
-- "Low-hanging fruit"
-
----
-
-## Humor and Personality
-
-- **Humor style:** Mostly absent in professional content. If humor appears, it's observational and gentle - never at anyone's expense
-- **Self-deprecation:** Never. Don't undermine your own credibility.
-- **Cultural references:** Avoid pop culture references. Stick to professional/work context.
-- **Analogies:** Use when helpful for explanation. Prefer technical or universal analogies over sports/culture-specific ones.
-
----
-
-## Transitions and Flow
-
-### How to Move Between Ideas
-
-- Use questions: "So what does this mean for..."
-- Use signposting: "Three things matter here..."
-- Use revelation: "Here's what I discovered..."
-- Use contrast: "The common approach is X. But what actually works is Y."
-
-### How to Conclude
-
-- Always tie back to practical implications
-- End with a specific action the reader can take
-- If no action possible, summarize the key insight clearly
-- Occasionally invite discussion, but don't overuse "What do you think?" as a crutch
-
----
-
-## Technical Depth Adaptation
-
-Match technical depth to the target audience:
-
-### For Leaders
-- High-level concepts
-- Business implications
-- Strategic decisions
-- ROI and outcomes
-- Avoid implementation details
-
-### For Low-Code Developers
-- Practical tips and patterns
-- Step-by-step guidance
-- Tool-specific insights
-- Common pitfalls and solutions
-- Can include some technical detail
-
-### For AI Architects
-- Technical depth welcome
-- Architecture patterns
-- Integration approaches
-- Trade-offs and decisions
-- Code snippets when relevant
-
-### For Power Users
-- Productivity gains
-- Workflow improvements
-- Tool comparisons
-- Time-saving techniques
-- Practical shortcuts
-
-**Key principle:** Always ensure technical content is well-explained and followable, regardless of depth. If you go technical, go all the way - don't half-explain.
-
----
 
 ## Language Guidelines
 
-- **Always English** for all LinkedIn content
+- Write in one consistent language per post; keep it accessible to non-native readers.
-- Clear, international English accessible to non-native speakers
+- Prefer simple sentence structures for complex ideas.
-- Avoid idioms that don't translate well internationally
-- Prefer simple sentence structures for complex ideas
-- Never use em dashes (—) - use hyphens, commas, or separate sentences instead
-
----
 
 ## Instructions for Claude
 
-When generating LinkedIn content for Kjell Tore:
+This is a placeholder. When it is in place (sentinel present), treat the defaults
-
+above as a reasonable baseline, and prompt the user to personalize via
-1. **Start with his voice profile** (from this document)
+`/linkedin:setup`. Once personalized, these instructions are replaced by the
-2. **Check the content pillar** - which audience is this for?
+user's own profile.
-3. **Choose appropriate storytelling technique** for the content type
-4. **Ensure actionable conclusion** - what can the reader DO?
-5. **Verify against Don'ts list** - no buzzwords, no criticism, no assumptions
-6. **Keep length in check** - 800-1500 characters for posts
-
-**Priority:** Sound like Kjell Tore > Optimize for algorithm
-
-**Exception:** If a phrase or approach would harm reach (external links, engagement bait), flag it but maintain his voice in everything else.
-
----
-
-## Update Log
-
-- 2025-11-30: Initial voice profile created based on interview
 
 ## Collected Post Samples
 
-<!-- Posts are saved here automatically by the Stop hook after each session where content is created. -->
+<!-- Posts are appended here automatically by the Stop hook after sessions where content is created. -->
 <!-- The voice-trainer agent uses these for 6-dimension drift scoring. Needs 5+ samples for reliable results. -->

plugins/linkedin-studio/assets/voice-samples/authentic-voice-samples.template.md

@@ -0,0 +1,61 @@
+# Authentic Voice Samples — Template
+
+Fill this in with YOUR voice, then save it as `authentic-voice-samples.md`
+(or, to keep it out of version control, `authentic-voice-samples.local.md`).
+`/linkedin:setup` can build this for you from 3–5 of your own posts.
+
+Delete every `[bracketed]` prompt as you replace it. Do NOT leave the
+`<!-- VOICE_PLACEHOLDER -->` sentinel anywhere in your finished profile — its
+presence keeps the voice personalization score at 0.
+
+## Core Voice Characteristics
+
+[List 4–6 traits that define how you write. For each: a one-line description and,
+where useful, the patterns you reach for. Examples: solution-oriented, factually
+grounded, story-led, non-judgmental, actionable.]
+
+## Do's
+
+- ✅ [Things that sound like you — openings, structure, language level, how you close.]
+
+## Don'ts
+
+- ❌ [Words, tones, and moves you avoid — buzzwords, em dashes, over-long posts, etc.]
+
+## Signature Phrases
+
+[The phrases you genuinely use to transition into insight or demonstration.
+Keep the list short and real — forced catchphrases read as fake.]
+
+## Vocabulary Preferences
+
+### Terms to always explain on first use
+[Domain acronyms/terms your audience may not know.]
+
+### Words/phrases to AVOID
+[Your personal no-go list.]
+
+## Language Guidelines
+
+- [Which language(s) you publish in, and any accessibility rules — e.g. simple
+  sentences for complex ideas, avoid hard-to-translate idioms.]
+
+## Technical Depth Adaptation
+
+[If you write for multiple audiences, note how depth shifts per audience
+(leaders vs practitioners vs power users).]
+
+## Instructions for Claude
+
+When generating content in this voice:
+1. Start from this profile.
+2. [Your priority order — e.g. "sound like me > optimize for algorithm".]
+3. Verify against the Don'ts list before finishing.
+
+## Update Log
+
+- [YYYY-MM-DD]: Initial voice profile created.
+
+## Collected Post Samples
+
+<!-- Posts are appended here automatically by the Stop hook after sessions where content is created. -->

plugins/linkedin-studio/commands/onboarding.md

@@ -139,7 +139,13 @@ Use AskUserQuestion:
 4. "Paste a paragraph you've written that sounds like YOU (email, doc, anything)"
 5. "Any words or phrases you'd NEVER use?"
 
-Save responses to `assets/voice-samples/authentic-voice-samples.md` under a new section `## Quick Voice Interview` (append, don't overwrite existing content).
+Save the responses to `assets/voice-samples/authentic-voice-samples.md`. **If the
+file is the shipped placeholder** (it contains `<!-- VOICE_PLACEHOLDER -->`),
+**REPLACE it entirely** with the profile built from the answers — the
+`<!-- VOICE_PLACEHOLDER -->` sentinel must NOT remain, or the voice score stays at
+0 after the user fills it in. **If the file is already a populated profile**, add a
+`## Quick Voice Interview` section instead of overwriting. Either way, the final
+file must contain no `<!-- VOICE_PLACEHOLDER -->`.
 
 **If user profile selected:** Ask for:
 1. Full name

plugins/linkedin-studio/commands/setup.md

@@ -25,7 +25,7 @@ Read these 8 asset files and detect placeholder patterns to calculate the curren
 
 | Category | Weight | File/Directory | Placeholder Detection |
 |----------|--------|----------------|----------------------|
-| Voice samples | 25 | `assets/voice-samples/authentic-voice-samples.md` | Check for `[Your Name]` or if file has <50 lines |
+| Voice samples | 25 | `assets/voice-samples/authentic-voice-samples.md` | Placeholder if it contains the `<!-- VOICE_PLACEHOLDER -->` sentinel (or has <50 lines) |
 | User profile | 20 | `config/user-profile.local.md` | Check if file exists; count `[Your ` placeholders |
 | Case studies | 15 | `assets/case-studies/*.md` | Count non-template `.md` files (exclude `case-study-template.md`) |
 | Frameworks | 10 | `assets/frameworks/*.md` | Count non-template `.md` files (exclude `framework-template.md`) |
@@ -96,14 +96,19 @@ Based on their answer, run the corresponding sub-workflow below.
    - How they handle technical depth
    - How they conclude (CTA style, takeaway style)
 4. Read the existing `assets/voice-samples/authentic-voice-samples.md`
-5. **Merge** new findings with existing content (don't overwrite existing data):
+5. **If the file is the shipped placeholder** (it contains `<!-- VOICE_PLACEHOLDER -->`):
+   **REPLACE it entirely** with the profile built from the user's samples. The
+   placeholder's `<!-- VOICE_PLACEHOLDER -->` sentinel must NOT survive — if it
+   does, the voice category stays at 0 even after the user fills in real data.
+   **Otherwise** (the file is already a populated profile), **merge** new findings
+   into the existing content (don't discard existing data):
    - Update "Core Voice Characteristics" if new patterns found
    - Add new entries to "Do's" and "Don'ts" lists
    - Update "Signature Phrases" with newly detected phrases
    - Add "Vocabulary Preferences" based on word analysis
    - Update "Update Log" with today's date
 
-6. Write the updated file back.
+6. Write the file back, and confirm it contains no `<!-- VOICE_PLACEHOLDER -->`.
 
 **Important:** Ask "Would you like to paste more samples?" after analyzing the first batch. More samples = better voice model.
 

plugins/linkedin-studio/hooks/scripts/__tests__/personalization-score.test.mjs

@@ -0,0 +1,66 @@
+import { describe, test, afterEach } from 'node:test';
+import assert from 'node:assert/strict';
+import { mkdtempSync, rmSync, mkdirSync, writeFileSync, existsSync } from 'node:fs';
+import { join } from 'node:path';
+import { tmpdir } from 'node:os';
+import { calculateScore } from '../personalization-score.mjs';
+
+// Step 5 (remediation): the shipped voice profile is a PII-free placeholder
+// carrying the sentinel below. Detection must key on the sentinel — NOT the old
+// `[Your Name]` heuristic — so the placeholder earns 0 of the 25 voice points,
+// and a populated profile (sentinel removed) earns them.
+const SENTINEL = '<!-- VOICE_PLACEHOLDER -->';
+
+// A >50-line body that contains NEITHER the sentinel NOR `[Your Name]`, so the
+// only thing distinguishing placeholder from real profile is the sentinel.
+const LONG_BODY = Array.from({ length: 60 }, (_, i) => `- voice characteristic line ${i + 1}`).join('\n');
+
+function makePluginRoot() {
+  const root = mkdtempSync(join(tmpdir(), 'lis-personalization-'));
+  mkdirSync(join(root, 'assets', 'voice-samples'), { recursive: true });
+  return root;
+}
+
+function writeVoice(root, content) {
+  writeFileSync(join(root, 'assets', 'voice-samples', 'authentic-voice-samples.md'), content, 'utf-8');
+}
+
+describe('calculateScore — voice samples (sentinel-based placeholder detection)', () => {
+  let root;
+
+  afterEach(() => {
+    if (root && existsSync(root)) {
+      rmSync(root, { recursive: true, force: true });
+    }
+    root = undefined;
+  });
+
+  test('placeholder with the sentinel scores 0 voice points (even >50 lines, no [Your Name])', () => {
+    root = makePluginRoot();
+    writeVoice(root, `# Voice Profile (placeholder)\n${SENTINEL}\n\n${LONG_BODY}\n`);
+
+    const { score, personalized } = calculateScore(root);
+
+    assert.equal(score, 0, 'placeholder must not earn the 25 voice points');
+    assert.equal(personalized, 0);
+  });
+
+  test('a real profile (no sentinel, >50 lines) earns the 25 voice points', () => {
+    root = makePluginRoot();
+    writeVoice(root, `# My Voice Profile\n\n${LONG_BODY}\n`);
+
+    const { score, personalized } = calculateScore(root);
+
+    assert.equal(score, 25, 'a populated profile must earn the 25 voice points');
+    assert.equal(personalized, 1);
+  });
+
+  test('a short placeholder (<50 lines) also scores 0 voice points', () => {
+    root = makePluginRoot();
+    writeVoice(root, `# Voice Profile (placeholder)\n${SENTINEL}\n`);
+
+    const { score } = calculateScore(root);
+
+    assert.equal(score, 0);
+  });
+});

plugins/linkedin-studio/hooks/scripts/personalization-score.mjs

@@ -16,11 +16,15 @@ export function calculateScore(pluginRoot) {
   const categories = 8;
 
   // --- 1. Voice samples (25 points) ---
+  // The shipped file is a PII-free placeholder carrying the VOICE_PLACEHOLDER
+  // sentinel. Key detection on the sentinel (deterministic) rather than the old
+  // `[Your Name]` heuristic: a populated profile removes the sentinel and earns
+  // the points; the placeholder (or any file still carrying it) scores 0.
   const voiceFile = join(pluginRoot, 'assets', 'voice-samples', 'authentic-voice-samples.md');
   if (existsSync(voiceFile)) {
     const content = readFileSync(voiceFile, 'utf-8');
     const lineCount = content.split('\n').length;
-    if (lineCount > 50 && !content.includes('[Your Name]')) {
+    if (lineCount > 50 && !content.includes('<!-- VOICE_PLACEHOLDER -->')) {
       score += 25;
       personalized += 1;
     }