x-grow

分类: 内容与多媒体 | 上传者: BlockRunAI | 下载: 0 | 版本: v1.0（最新）

Use when crafting X posts, optimizing drafts, or generating post images - helps write high-performing content based on X's ranking algorithm

更新日志: Source: GitHub https://github.com/BlockRunAI/x-grow

目录结构

当前层级: 根目录

📁 docs/
- 📁 plans/
  - 📄 2026-01-20-x-grow-design.md 9.0 KB
📁 knowledge/
- 📄 algorithm.md 2.8 KB
- 📄 best-practices.md 3.2 KB
📁 prompts/
- 📄 draft.md 2.0 KB
- 📄 image.md 1.9 KB
- 📄 review.md 2.3 KB
📄 README.md 4.4 KB
📄 SKILL.md 6.3 KB
📄 smithery.yaml 429 B

SKILL.md

---
name: x-grow
description: Use when crafting X posts, optimizing drafts, or generating post images - helps write high-performing content based on X's ranking algorithm
---

# X Grow

## Overview

Help users write high-performing X posts by combining algorithm knowledge, best practices, and optional real-time research via BlockRun MCP.

**Three modes:**
- `/x-grow draft "topic"` - Research and write optimized posts
- `/x-grow review` - Analyze and improve user's draft
- `/x-grow image "description"` - Generate post visuals

## Mode Detection

```dot
digraph mode_detection {
    "User invokes /x-grow" [shape=ellipse];
    "Has 'draft' + topic?" [shape=diamond];
    "Has 'review'?" [shape=diamond];
    "Has 'image'?" [shape=diamond];
    "DRAFT MODE" [shape=box];
    "REVIEW MODE" [shape=box];
    "IMAGE MODE" [shape=box];
    "Ask which mode" [shape=box];

    "User invokes /x-grow" -> "Has 'draft' + topic?";
    "Has 'draft' + topic?" -> "DRAFT MODE" [label="yes"];
    "Has 'draft' + topic?" -> "Has 'review'?" [label="no"];
    "Has 'review'?" -> "REVIEW MODE" [label="yes"];
    "Has 'review'?" -> "Has 'image'?" [label="no"];
    "Has 'image'?" -> "IMAGE MODE" [label="yes"];
    "Has 'image'?" -> "Ask which mode" [label="no"];
}
```

## BlockRun MCP Detection

Before drafting, check if BlockRun MCP is available:

```dot
digraph blockrun_check {
    "Starting draft/research" [shape=ellipse];
    "BlockRun MCP available?" [shape=diamond];
    "Use blockrun_twitter for research" [shape=box];
    "Use baked-in knowledge only" [shape=box];
    "Continue to generation" [shape=ellipse];

    "Starting draft/research" -> "BlockRun MCP available?";
    "BlockRun MCP available?" -> "Use blockrun_twitter for research" [label="yes"];
    "BlockRun MCP available?" -> "Use baked-in knowledge only" [label="no"];
    "Use blockrun_twitter for research" -> "Continue to generation";
    "Use baked-in knowledge only" -> "Continue to generation";
}
```

**To check:** Look for `blockrun_twitter` or `blockrun_chat` tools in available MCP tools.

## Draft Mode

When user runs `/x-grow draft "topic"`:

### Step 1: Research (if BlockRun available)

Use `blockrun_twitter` to search:
```
blockrun_twitter query: "high performing posts about [topic]"
```

Find 3-5 examples, identify what's working.

### Step 2: Apply Algorithm Knowledge

Reference `knowledge/algorithm.md` for:
- Engagement weights (replies = +13.5, author reply = +75)
- Timing importance (first 10 min critical)
- Negative signals to avoid

Reference `knowledge/best-practices.md` for:
- Hook patterns (curiosity gap, contrarian, specific numbers)
- Post structure
- Engagement triggers

### Step 3: Generate Output

```
## POST (ready to copy):

[The optimized post - full text]

---

## VARIATIONS:

**1. Different hook:**
[Alternative opening angle]

**2. Thread opener:**
[If topic warrants a thread]

**3. Casual tone:**
[More conversational version]

---

## WHY THIS WORKS:

- **Hook:** [Explain the hook strategy used]
- **Algorithm:** [Which ranking factors it optimizes for]
- **Format:** [Why this structure was chosen]

---

## BEST PRACTICES APPLIED:

- [Specific practice 1]
- [Specific practice 2]
- [Specific practice 3]

---

## IMAGE SUGGESTION:

This post would benefit from a visual.
Type `/x-grow image "[suggested description]"` to generate one.
```

## Review Mode

When user runs `/x-grow review` and provides their draft:

### Step 1: Score Against Algorithm

Evaluate each factor:
- Hook strength (1-10)
- Engagement trigger presence
- Length optimization
- Negative signal check
- Format effectiveness

### Step 2: Generate Output

```
## SCORE: [X]/10

---

## CHECKLIST AUDIT:

[Use checkmarks, X marks, and warning signs]

| Factor | Status | Notes |
|--------|--------|-------|
| Hook | [status] | [specific feedback] |
| Length | [status] | [specific feedback] |
| Engagement trigger | [status] | [specific feedback] |
| Specificity | [status] | [specific feedback] |
| Format | [status] | [specific feedback] |
| Negative signals | [status] | [specific feedback] |

---

## OPTIMIZED VERSION:

[Rewritten post incorporating improvements]

---

## WHAT CHANGED:

- [Change 1 and why]
- [Change 2 and why]
- [Change 3 and why]

---

## BEST PRACTICE TIP:

[One actionable insight about what they can improve generally]
```

## Image Mode

When user runs `/x-grow image "description"`:

### Check BlockRun Availability

Image generation requires BlockRun MCP. If not available:
```
Image generation requires BlockRun MCP with a funded wallet.

Install: claude mcp add blockrun -- npx @blockrun/mcp
Fund wallet with USDC on Base network.

Cost: ~$0.05 per image via Nano Banana
```

### Generate with BlockRun

Use `blockrun_image`:
```
blockrun_image prompt: "[description optimized for X visuals]" model: "google/nano-banana"
```

### Output

```
## GENERATED IMAGE:

[Display/reference the generated image]

---

## IMAGE OPTIMIZATION TIPS:

- High contrast stops the scroll
- Minimal text (algorithm prefers native images)
- Sized 1200x675 for optimal X preview
- Posts with images get 2x engagement

---

## SUGGESTED POST:

[Draft a post to accompany this image]
```

## On-Demand Research

User can ask for research without drafting:

`/x-grow research "topic"` or just ask "what's working for [topic]?"

Use `blockrun_twitter` to find:
- Recent high-performing posts
- Common hooks/formats
- Engagement patterns

Present findings without drafting.

## Quick Reference

### Algorithm Weights (memorize these)
| Action | Weight |
|--------|--------|
| Author replies to comment | +75.0 |
| Get replies | +13.5 |
| Profile clicks | +12.0 |
| Click-through | +11.0 |
| Retweet | +1.0 |
| Like | +0.5 |
| "Show less" | -74.0 |
| Report | -369.0 |

### Hook Patterns
- Curiosity gap: Promise value, deliver after engagement
- Contrarian: Challenge conventional wisdom
- Specific numbers: Concrete beats vague
- Problem statement: Lead with pain
- Bold claim: Statement worth engaging with

### Red Flags
- "I'm excited to announce..."
- 4+ hashtags
- ALL CAPS
- Asking for likes/RTs explicitly
- No engagement trigger

## Cost Summary (with BlockRun)

| Action | Cost |
|--------|------|
| Draft with research | ~$0.03 |
| Draft with image | ~$0.08 |
| Image only | ~$0.05 |
| Review only | Free |
| Research only | ~$0.03 |

## Sources

- [X Algorithm Repo](https://github.com/xai-org/x-algorithm)
- [BlockRun MCP](https://github.com/BlockRunAI/blockrun-mcp)

登录后下载/点赞/收藏 ❤ 7 | ★ 0

x-grow

目录结构

SKILL.md

举报内容

提示