Token导航 LogoToken导航

Agent Skill · 研究检索

build-clusters(建立集群)

build-clusters 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。

build-clusters 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为调用外部服务;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 3,881。

总安装 38,677周安装 1,585GitHub Stars 7下载 12,482核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:build-clusters(建立集群)
来源仓库:https://github.com/calm-north/seojuice-skills
仓库路径:skills/build-clusters
安装命令:
npx skills add https://github.com/calm-north/seojuice-skills --skill build-clusters
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/calm-north/seojuice-skills --skill build-clusters

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

通过支柱架构和互联策略构建主题权威集群。

根据搜索数据、意图模式和受众研究生成 8-20 个子主题,以围绕种子关键字填充集群 具有综合支柱页面(中心)和重点分支页面(卫星)的结构集群,具有强制双向链接和同级页面之间的横向交叉链接 对现有内容覆盖率进行评分,识别差距(全部、部分或无),并标记需要创建或刷新的页面 提供生产排序、具有锚文本指导的互连表以及集群健康指标(覆盖范围、链接健康状况、内容质量),以确定排名的准备情况

来源仓库内路径:skills/build-clusters

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/calm-north/seojuice-skills --skill build-clust…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要调用外部服务,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

build-clusters 怎么安装?

在终端执行 npx skills add https://github.com/calm-north/seojuice-skills --skill build-clusters。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

build-clusters 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

build-clusters 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

build-clusters 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

build-clusters 用的人多吗?

公开数据是累计安装 38,677 次,近一周安装 1,585 次,下载 12,482 次,GitHub 7 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

build-clusters 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
build-clusters

建立集群 · 本页

38,6777调用外部服务未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Build Clusters

Build a topical authority cluster from a seed keyword using pillar-spoke structure, coverage scoring, and interlinking plans.

What is a Topic Cluster?

A topic cluster is a group of interlinked pages that collectively cover a subject area. Google evaluates topical authority at the cluster level — ranking a single page is harder if the site has no supporting content around the topic.

Structure:

  • Pillar page — comprehensive overview of the broad topic (targets head term)
  • Spoke pages — focused articles covering subtopics (target body/long-tail terms)
  • Internal links — every spoke links to the pillar, pillar links to all spokes, spokes cross-link to siblings

Before You Start

Gather this context (ask if not provided):

  1. Seed topic. The broad subject area to build authority around.
  2. Existing content. Does the site already have pages on this topic? List them.
  3. Business relevance. How does this topic connect to the product or service?
  4. Content capacity. How many pieces can the team produce per month?

Step 1: Subtopic Discovery

From the seed topic, generate subtopics using these methods:

Search-derived:

  • People Also Ask questions for the seed keyword
  • Related searches at the bottom of SERPs
  • Autocomplete suggestions (seed + a, b, c...)
  • Competitor content analysis — what subtopics do top-ranking sites cover?

Intent-derived:

  • Awareness: "what is [topic]", "why [topic] matters"
  • Consideration: "best [topic] tools", "[topic] vs [alternative]"
  • Implementation: "how to [topic]", "[topic] tutorial"
  • Troubleshooting: "[topic] not working", "common [topic] mistakes"

Audience-derived:

  • Beginner questions about the topic
  • Advanced practitioner concerns
  • Decision-maker evaluation criteria

Aim for 8-20 subtopics per cluster.

Step 2: Cluster Map

Organize subtopics into a structured cluster:

Pillar: [Broad Topic] (head term)
│
├── Spoke: [Subtopic 1] (body term)
│   └── Intent: informational
│
├── Spoke: [Subtopic 2] (body term)
│   └── Intent: commercial investigation
│
├── Spoke: [Subtopic 3] (long-tail)
│   └── Intent: transactional
│
├── Spoke: [Subtopic 4] (long-tail)
│   └── Intent: informational
│
└── ... (8-15 more spokes)

Step 3: Coverage Scoring

Score how well the existing site covers the cluster:

SubtopicExisting Page?Quality (1-5)TrafficGap?
[subtopic 1]/blog/topic-14500/moNo
[subtopic 2]———Yes
[subtopic 3]/blog/old-post250/moPartial (needs refresh)
  • Full gap — no existing page, needs creation
  • Partial gap — page exists but is thin, outdated, or off-intent
  • Covered — strong existing page, may just need internal linking

Step 4: Pillar Page Design

The pillar page should:

  • Cover the topic comprehensively at an overview level (2,000-4,000 words)
  • Link to every spoke page for deeper dives
  • Be structured as a table of contents for the entire cluster
  • Target the highest-volume keyword in the cluster
  • Include a summary of each subtopic (2-3 paragraphs) with a link to the full spoke

Pillar page is NOT a mega-article that tries to cover everything in depth. It is a hub that distributes authority and directs readers to the right spoke.

Step 5: Interlinking Plan

Map the internal links:

From PageTo PageAnchor TextContext
PillarSpoke 1"[subtopic 1] guide"In the subtopic 1 overview section
Spoke 1Pillar"[broad topic]"In the introduction or conclusion
Spoke 1Spoke 2"[subtopic 2]"Where subtopic 2 is mentioned contextually
Spoke 3Spoke 1"[subtopic 1]"Where comparison is relevant

Rules:

  • Every spoke links to the pillar (mandatory)
  • Pillar links to every spoke (mandatory)
  • Spokes cross-link to 2-4 siblings (where contextually natural)
  • Use varied anchor text (not always the exact keyword)

Step 6: Cluster Health Metrics

Score the cluster's readiness to compete:

Coverage Score  = (Covered spokes / Total spokes) x 100
Link Health     = (Spokes with bidirectional pillar link / Total spokes) x 100
Content Quality = (Spokes scoring 3+ quality / Total spokes) x 100
MetricScoreThreshold
Coverage[x]%> 70% to start ranking for pillar keyword
Link Health[x]%100% is the target — every spoke must link to pillar and back
Content Quality[x]%> 80% — clusters with thin spokes dilute authority

Expected Outcomes by Health Level

Cluster StateCoverageLink HealthContent QualityExpected Result
Incomplete<50%<70%AnyPillar unlikely to rank page 1; spokes rank individually at best
Developing50-70%70-90%50-80%Pillar may appear page 2-3; some spokes rank for long-tail
Competitive70-90%100%80-90%Pillar competes for page 1; most spokes rank for their targets
Dominant>90%100%>90%Pillar strong on page 1; cluster captures most queries in the topic

Internal Link Standards per Cluster

Link TypeMinimum CountDirection
Pillar → each spoke1 per spokeDownward — link from the relevant section of the pillar
Each spoke → pillar1 per spokeUpward — "our complete [topic] guide"
Spoke ↔ sibling spokes2-4 per spokeLateral — where contextually natural
Cross-cluster bridges0-2 per clusterBetween hubs — only with genuine topical relevance

Cluster ready to compete: Coverage > 70%, Link Health = 100%, Content Quality > 80%. Below these thresholds, prioritize filling gaps before expecting the pillar to rank.

Step 7: Production Sequence

Order the content production for maximum impact:

  1. Pillar page first — even as a draft, it establishes the hub
  2. Highest-opportunity spokes next — pages targeting gaps with the best opportunity scores
  3. Refresh existing spokes — update and relink any partial-gap pages
  4. Remaining spokes — fill out the cluster over time
  5. Update pillar — add links to each new spoke as it's published

Output Format

Topic Cluster: [seed topic]

Cluster Summary

  • Pillar keyword: [keyword] (volume: [x], difficulty: [y])
  • Total spokes: [count]
  • Existing coverage: [x]% ([n] pages exist, [n] need creation)
  • Total cluster volume: [sum of all keyword volumes]

Cluster Map [Visual structure from Step 2]

Coverage Scorecard [Table from Step 3]

Pillar Page Spec

  • Target keyword: [keyword]
  • Recommended title: [title]
  • Structure: [heading outline with spoke links]

Interlinking Plan [Table from Step 5]

Production Roadmap [Ordered list from Step 6 with estimated timelines]


Pro Tip: Use the free Blog Keyword Generator to discover subtopics for your cluster. SEOJuice MCP users get automatic cluster mapping — run /seojuice:content-strategy to see existing clusters with coverage metrics, or use list_clusters and get_cluster_detail to check cluster health and identify gaps.

平台分布

Codex

36.63%
按下载量换算4,572

Claude

29.38%
按下载量换算3,667

Cursor

15.91%
按下载量换算1,986

Gemini CLI

9.34%
按下载量换算1,166

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills