Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问许可证需确认审计通过

algorithm-mechanism算法机制

Agent Skill

algorithm-mechanism 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

1,582

周安装

64

GitHub Stars

142

下载量

497
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:algorithm-mechanism(算法机制)
来源仓库:https://github.com/vivy-yi/xiaohongshu-skills
仓库路径:skills/algorithm-mechanism
安装命令:
npx skills add https://github.com/vivy-yi/xiaohongshu-skills --skill algorithm-mechanism
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/vivy-yi/xiaohongshu-skills --skill algorithm-mechanism

简介

algorithm-mechanism 解析小红书推荐机制,指导内容曝光与互动优化策略。

  • 适用于新账号冷启动、爆款内容复盘或流量波动诊断的场景。
  • 聚焦算法权重因子如完播率、互动密度与标签准确性,而非单纯追求发布频率。
  • 安装方式基于 GitHub,使用前应确认是否支持平台数据接口调用。
  • 不替代原创创意,仅提供机制层面的执行建议与调优方向。

SKILL.md

Algorithm Mechanism (算法机制)

Overview

Xiaohongshu's recommendation algorithm determines which content gets shown to whom. Understanding how it works enables you to optimize content for maximum reach and engagement rather than posting blindly.

When to Use

Use when:

  • Views are consistently low (<100 for new accounts)
  • Content performance is unpredictable
  • Want to increase content discovery
  • Some posts go viral while others flop
  • Planning content optimization strategy
  • Troubleshooting reach issues

Do NOT use when:

  • Just need to post content (use graphic-content-creation)
  • Need real-time trend analysis (use qiangua-data)

Core Pattern

Before (algorithm blindness):

❌ "I'll post when I feel like it"
❌ "Why did this post get only 50 views?"
❌ "The algorithm just doesn't like me"

After (algorithm-aware):

✅ Post at peak hours when my audience is most active
✅ Analyze why similar posts got 10x more views
✅ Optimize title, cover, and hashtags for algorithm signals

5 Algorithm Factors:

  1. Initial Engagement - First 30 minutes determines distribution
  2. Content Quality - Completion rate, saves, comments
  3. Account Authority - Consistency, expertise, community trust
  4. User Interaction - Past engagement with similar content
  5. Timeliness - Recency, trending topics, seasonal relevance

Quick Reference

FactorWhat It MeasuresOptimization Strategy
Initial engagementFirst 30min performancePost at peak hours, engage immediately
Completion rateHow many finish reading/watchingKeep content concise, add value early
Save rateContent worth revisitingInclude actionable tips, resources
Comment rateDiscussion qualityAsk questions, respond to comments
Share rateShareabilityRelatable, valuable content
Follow rateAccount appealConsistent positioning, clear value proposition

Implementation

Step 1: Understand the 3-Stage Distribution

Stage 1: Initial Testing (0-30 minutes)

  • Algorithm shows content to 200-500 initial users
  • Measures initial engagement (likes, comments, shares, completion)
  • Critical: Performance here determines whether content advances

Stage 2: Expanded Distribution (30min - 24 hours)

  • If initial engagement is good, expands to 5,000-20,000 users
  • Continues measuring engagement signals
  • Good performance → Stage 3, poor performance → stops here

Stage 3: Viral Distribution (24 hours - 7 days)

  • Top performers reach 50,000-1,000,000+ users
  • Appears in "Recommended" and "Explore" feeds
  • Can continue getting views for weeks

Step 2: Optimize for Initial Engagement

The "First 30 Minutes" Strategy:

Pre-Publishing (1 hour before posting):

  • Research trending hashtags in your niche
  • Analyze top 5 posts similar to your planned content
  • Optimize title for curiosity and relevance
  • Ensure cover image is high-quality and eye-catching
  • Write compelling first sentence (hook)

Publishing (choose optimal time):

  • Best times: 7-9pm (weekday evenings), 9-11am (weekends)
  • Your best time: Check your audience data for peak hours
  • Avoid: 2-5pm (low activity), 11pm-7am (most users asleep)

First 30 Minutes (critical window):

  • Stay active, respond to every comment immediately
  • Share to your other social platforms (WeChat, etc.)
  • Ask a question in the post to encourage comments
  • Engage with similar creators' content (build reciprocity)

Step 3: Optimize Content for Algorithm Signals

Title Optimization:

✅ Use numbers: "5个..."
✅ Ask questions: "你知道...吗?"
✅ Create curiosity: "没想到..."
✅ Include keywords: "[vertical] + [benefit]"
❌ Avoid: Too generic, no clear benefit

Cover Optimization:

  • High contrast: Bright colors, clear text overlay
  • Text overlay: 3-5 keywords maximum
  • Face/photo quality: Clear, well-lit, professional
  • Aspect ratio: 3:4 or 1:1 (best for feed)

Hashtag Optimization:

  • Use 3-5 relevant hashtags (not more)
  • Mix: 1 broad (#穿搭), 2-3 specific (#韩系穿搭 #小个子), 1 niche (#平价穿搭)
  • Research trending tags in your vertical

Content Structure:

Hook (first 3 seconds) → Value (main content) → CTA (ask for engagement)

Example:
Hook: "这个方法让我在1个月涨了5000粉" (curiosity)
Value: [分享具体方法步骤] (actionable)
CTA: "你有什么涨粉秘诀?评论区告诉我" (engagement)

Step 4: Build Account Authority

Consistency signals:

  • Post 3-4 times per week (not sporadic)
  • Maintain consistent content theme (vertical focus)
  • Respond to 80%+ of comments within 24 hours
  • Keep quality high (delete low-performing posts)

Expertise signals:

  • Demonstrate knowledge in your niche
  • Share original insights, not just reposts
  • Use data and evidence to support claims
  • Engage with other experts in your field

Community trust signals:

  • Authentic engagement (not just self-promotion)
  • Help others (answer questions, provide value)
  • Transparent about sponsorships (builds trust)
  • Positive interactions (no drama or controversy)

Step 5: Analyze and Iterate

Weekly Algorithm Performance Review:

Track these metrics:

  • Which posts reached Stage 2/3 distribution?
  • What common traits do viral posts share?
  • What time posted? What hashtags used?
  • What was initial engagement rate?

Identify patterns:

  • "Posts with numbers in titles get 2x more views"
  • "Skincare tutorial videos outperform image posts"
  • "Posts at 8pm consistently outperform 2pm"

Double down on what works:

  • Create more content similar to top performers
  • Post at times that consistently work
  • Use formats that achieve wide distribution

Common Mistakes

MistakeWhy HappensFix
Posting at random timesConvenienceSchedule for peak hours (7-9pm)
Ignoring first 30 minutesDon't know importanceEngage intensely in first 30min
Too many hashtagsMore is better mythUse 3-5 relevant hashtags
Inconsistent postingBusy scheduleBatch create and schedule
Deleting low-performing posts too quicklyEmbarrassedWait 7 days, analyze why it failed

Real-World Impact

Algorithm-aware accounts (real examples):

  • Account A: Posted at random times → 3x view increase after shifting to 8pm posting
  • Account B: Generic titles → 5x engagement increase with curiosity-driven titles
  • Account C: Ignored first 30min → 10x distribution by engaging immediately after posting

Key insight: Understanding and optimizing for the algorithm can increase average views by 3-10x.


Related Skills:

  • REQUIRED: data-analytics (measure algorithm performance)
  • REQUIRED: viral-strategy (create algorithm-friendly viral content)
  • REQUIRED: account-weight (understand account authority)
  • hashtag-optimization (optimize hashtag strategy)
  • traffic-analysis (analyze where traffic comes from)

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

34.17%
按下载量换算170

Claude

28.96%
按下载量换算144

Cursor

19.61%
按下载量换算97

Gemini CLI

9.99%
按下载量换算50

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills