write-xiaohongshu

Write Xiaohongshu（小红书：先研究再写再发）

Quick Start（最短用法）

When the user gives you a topic, run the full process "Step 0→7"; final output only:

• Title (≤20 characters)
• Body (≤1000 characters)
• Tags (#xxx #yyy …)
• Images (use user-provided images if available, otherwise select 1–2 images, 9:16 preferred)
• Publish after user review
• Publish result (if Xiaohongshu MCP is available)

Hard Limits (Must Follow)

• Title: ≤ 20 characters
• Body: ≤ 1000 characters

Counting method: count by "characters" (including spaces, punctuation, line breaks, #tags, etc.). To be safe: title ≤18, body ≤950, leaving room for tags/formatting.

Output (Must Include These Blocks)

This skill's output is divided into three sections:

1. Analysis Summary Report (must be given after completing Step 1+2 analysis, structure as template below)
2. Final Publishable Draft (title + body + tags + images)
3. Publish result after review

Title: <no more than 20 characters>
Body:
<no more than 1000 characters>
Tags: #tag1 #tag2 #tag3 ...
Images: <1~2 images, 9:16 preferred; provide image URL or local path>

If the publish API supports a separate "tags" field, do not pile #tags in the body; put tags in the publish parameters; still ensure the body itself ≤1000.

Full Process (Must Follow Order)

flowchart TD
  A[Check login status] --> B[Search Xiaohongshu for Top10 image posts]
  B --> C[Analyze title & body patterns + reasons for high likes]
  C --> D[Fetch comments and analyze emotional resonance points]
  D --> E[Firecrawl to supplement background & fact-check]
  E --> F[Write title/body/tags + character count self-check]
  F --> G{User provided images?}
  G -- Yes --> H[Use user images]
  G -- No --> I[Find 1~2 high-quality 9:16 images online]
  H --> J[Show content to user for confirmation]
  I --> J
  J --> K{User confirmed?}
  K -- Confirmed --> L[Publish via Xiaohongshu MCP]
  K -- Needs modification --> F

Step 0: Check Login Status First

Call the Xiaohongshu MCP to check login status; if not logged in, get a QR code for the user to log in.

Step 1: Search Xiaohongshu for Related "Top Likes" Image Posts and Summarize Patterns

Goal: First find the most popular writing styles among similar content, then write.

Hard Requirement (Must Execute)

• Must successfully obtain at least 10 "post details fully usable" image posts before proceeding to Steps 2/3/4/5/6/7.

Requirements:

• Only select image posts (exclude video/live clips)
• Prioritize relevance, then high like count, filter by "most likes"
• Filter by "within half a year"
• Only use the Xiaohongshu MCP's get_specified_post tool to fetch, filter for "image posts", "within half a year", get 20 posts (actual results may be fewer than 20, tolerance 10), do not call other MCP tools; the tool returns all needed information.

For each top post, record these fields (for subsequent analysis):

• Title (original)
• Likes/saves/comments (if available)
• Body structure (summarize in 1 line: e.g., "conclusion → list → steps → feedback → interaction")
• Key hook (the "attraction point" of the first sentence)
• Evidence type (experience, comparison, data, screenshot, before/after…)
• Interaction design (questions, polls, comment keywords, giveaways, collection guides…)

Output a "pattern summary" (directly reusable for writing):

• Title patterns (length, sentence structure, keywords, symbols/numbers, emotional words, benefit points)
• Body patterns (paragraph rhythm, lists/steps, twists, pitfalls, ending CTA, tone style)
• Reasons for high likes (at least 3, as concrete as possible: "writing style/information density/resonance points/actionability/tone")

Step 2: Analyze "Emotional Resonance" Reasons Based on Comments from Step 1 Data

Goal: Find out why the audience "wants to comment/share/save", summarize 3-6 resonance point conclusions, and 1 interaction design suggestion.

Method:

• Categorize "resonance triggers" in comments, e.g.:
  • Shared pain points (Me too…)
  • Identity (office worker/student/mom…)
  • Anxiety and solution (Finally a way/Saved)
  • Values (self-discipline, relaxed vibe, value for money, minimalism…)
  • Contrast/surprise (So that works too)
  • Replicability (I can do it too)
  • Sharing desire (I want to try too)
  • Curiosity (What is this?)
  • Resonance point (I feel the same)

Output:

• 3～6 "resonance point conclusions" (each with 1～2 original quote fragments/summaries)

Analysis Summary Report (Must Output, Then Automatically Continue Subsequent Process)

After completing Steps 1+2 (and meeting the "usable samples ≥10" condition), you must first output a summary report following the structure below (content style/granularity refer to user-provided examples), then in the same round of output automatically continue Steps 3→7.

If the sample threshold is not met: stop immediately per Step 1's hard requirement, do not proceed to Steps 2/3/4/5/6/7, and do not fabricate the report.

Report template:

【Analysis Summary Report】
Comment count/interaction ranking (high to low)
1. <title> - <comment count> (<image/video>)
2. ...
...
10. ...
11. ...

Title pattern analysis
- <pattern 1>
- <pattern 2>
- <pattern 3>
...

Content pattern analysis
- <pattern 1>
- <pattern 2>
...

Cover pattern analysis
- <pattern 1>
- <pattern 2>
...

Summary
Three key elements for viral posts:
1. <element 1>
2. <element 2>
3. <element 3>

Interaction catalysts:
- <catalyst 1>
- <catalyst 2>
...

Step 3: Use Firecrawl MCP to Supplement Background Knowledge (and Fact-Check)

Goal: Avoid empty content, avoid errors/exaggeration, fill in "credible information".

Method:

• Use topic keywords for expanded search: concept explanations, common misconceptions, authoritative advice, data (preferably from credible sources)
• Refine into "5～10 background key points that can be cited" (short sentences)
• Downgrade claims that are easily questioned (efficacy, benefits, comparative conclusions):
  • Change to "my personal experience/suitable for some people/may work"
  • Or add conditions (target audience, prerequisites)

Output a "background knowledge card":

• Key conclusions (5～10 items)
• Risks/notes (2～4 items)

Step 4: Write Title + Body + Tags Based on Above Analysis (and Enforce Character Count Self-Check)

Writing requirements (Xiaohongshu-style, but don't overdo it):

• Conversational, short sentences, line breaks every 1–3 sentences
• High information density: less fluff, more "actionable steps/lists"
• Use emojis appropriately (to enhance rhythm/emotion), but don't pile them
• End naturally, like a real user wrote it, not "marketing tone"

De-AI Prompt (Must Execute)

• Write more "I/I at that time/I found/the pitfalls I stepped into", use less "everyone/users/audience/suggestions as follows"
• Give 2～3 specific details: time point, scene, before/after comparison, a specific line/a shot/a small action
• Allow imperfection: use "I feel/I guess/maybe/not sure but…" (don't act authoritative)
• Less template language: avoid "in summary/to sum up/therefore/first second finally/not to be missed/quickly"
• Like chatting with a friend: short sentences, line breaks, occasional colloquial words (but don't pile)

Character count gate (must do at the end):

• Title character count > 20: delete redundant words, remove subtitles, shorten sentences, remove unnecessary symbols
• Body character count > 1000: first delete repetitions and filler words, then merge sentences, then compress steps/reduce examples
• Final output only the compliant version

Step 5: If User Didn't Provide Images, Find 1～2 High-Quality Images Online (9:16 Preferred)

Requirements:

• 1～2 images only, high quality, clean, strongly related to the topic
• Aspect ratio priority 9:16 (portrait)
• Only select commercial use/royalty-free sources (e.g., Pexels / Unsplash, etc.)
• Output: image URL (and optionally: local download path)

Step 6: Show Content to User for Confirmation (Review)

Goal: Let the user confirm the content before publishing, ensuring it meets expectations.

Must Execute:

After completing Steps 4+5, you must first show the complete content to the user for confirmation, format as follows:

【Preview of Content to Be Published】

Title: <title content> (<character count>/20)
Body:
<body content>
(<character count>/1000)

Tags: #tag1 #tag2 #tag3 ...

Images:
- Image 1: <URL or path>
- Image 2: <URL or path> (if any)

---
Please confirm:
- Enter "confirm" or "publish": proceed to publish
- Enter "modify" or specific modification suggestions: adjust content based on your feedback
- Enter "cancel": terminate the publish process

Wait for user confirmation:

• If user confirms: proceed to Step 7 to publish
• If user requests modification: modify content based on user feedback, re-display for confirmation (back to Step 6)
• If user cancels: terminate process, do not publish

Note: Do not automatically proceed to publish step until the user explicitly confirms.

Step 7: Publish via Xiaohongshu MCP

When publishing (adapt to MCP capabilities, do as much as possible):

• Upload 1～2 images (portrait preferred)
• Set title, body, tags
• If supported: create draft first → display draft info → then publish

After successful publish, output:

• Published link/ID (if available)
• Publish parameter summary (title/tags/image count)

Fallback Strategy (When Tools Unavailable)

If the current environment lacks Firecrawl MCP or Xiaohongshu MCP:

• Still complete the process "analysis → writing → image links", but do not auto-publish
• List the information the user needs to supplement as a minimal checklist (e.g., provide 3~5 reference post links, or authorization/interface configuration)

Examples

Example 1: User Only Gives Topic

User input:

Write a Xiaohongshu image post: topic is "quick dinner for office workers", suitable for one person, budget under 20 yuan.

You should:

• Use topic keywords to search Xiaohongshu for "image posts", summarize title/body patterns
• Fetch comments from these posts, extract resonance points and interaction questions
• Use Firecrawl to supplement "health/meal prep/safety" background
• Output title + body + tags, and do character count self-check
• If no images, find 1~2 high-quality 9:16 portrait images
• Show content to user for confirmation
• After user confirms, publish via Xiaohongshu MCP (if available)

Example 2: User Provides Images

User input:

I have 5 portrait images (9:16), help me write a Xiaohongshu post, topic is "makeup pitfalls for beginners", gentle tone preferred.

You should:

• Still first do the Top10 + comments + background three-step analysis
• After writing the copy, show content to user for confirmation
• After user confirms, publish using the user-provided images