A
abduladnan-mt7/Mt7-Seo-king
Evidence-tiered SEO skill for Claude Code: Google's docs first, data second, community tactics only when…
在 GitHub 查看 ↗链接已于 2026-11-12 核查 · HTTP 200 · 打开的是厂商自己的官网
- ★ 数取值
- 1 ★ · abduladnan-mt7/Mt7-Seo-king · 取自 GitHub 搜索 API · (2026-11-13 20:00:50 核对)
- 来源明细
- 从该仓库的 GitHub 页面直接读取 · 条目类型:开源项目
- 收录与活跃
- 收录于 2026-10-29 · 最近见到 2026-11-14 · 共见到 17 次 · 被 4 个来源列出
- 评分构成
- 综合分 403 · 1 ★ · 4 个来源 · repo
常见问题(答案都来自我们的记录)
它适合谁 / 用来做什么?
我们把它归在这些标签下: Agent 技能。标签来自条目自身,不是我们强加的分类。
谁维护它、还在更新吗?
仓库
abduladnan-mt7/Mt7-Seo-king;我们收录于 2026-10-29;最近见到 2026-11-14。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。# Mt7 SEO King
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
An SEO skill for Claude Code (and any agent that reads Agent Skills) that tells you what actually moves rankings, and how sure it is.
Every recommendation carries an evidence tag. [G] means Google's own documentation, [DATA] a study or measured test with its sample size and date, and [PROVEN] independent practitioners agreeing. [CLAIM], [WARNED] and [INFERENCE] are labelled as such, so you can tell a rule from a hunch. Where community advice contradicts Google, the skill follows Google and says so. Nothing paid is required: no API keys, and nothing to install beyond Node 18 for the scripts. If you use OpenSEO, the skill can also draw on its live keyword, SERP and backlink data, and asks before any call that spends credits.
## What it does
Audits and diagnosis. A phased workflow (diagnose → foundations → intent map → pages → local and reviews → links → AI search → measure). Every phase has a gate to pass before moving on.
Diagnose mode. diagnose.mjs crawls a site the way a polite, non-rendering crawler does. It ranks what it finds from critical to info, and gives each finding its per-URL evidence, its Google source and a way to verify it. On crawler-test.com, a site built with known SEO faults, four runs found 160 of its 164 test issues.
Traffic drops. A dated ledger of Google core, spam and policy updates, each entry citing a Google URL, plus Search Console's own logging errors. --around shows what overlapped a drop before anyone blames "the algorithm".
Verify before you report. Audit rules that stop the usual false positives: "your sitemap lists noindexed pages" (it gets checked first), "no schema" (on a JS-rendered page), "duplicate Organization" (a nested parent organization isn't one).
Pages that rank without becoming spam. Templates for service, industry, city, case-study, comparison and how-to-choose pages, each with a legitimacy gate. A programmatic-SEO gate covers tools, conversions and comparisons built from one template. Each page type also has its "becomes a doorway page when..." line.
Links at no cost. A vetted catalogue of free link and mention sources with their link attributes, outreach kits (lists, digital PR, journalist sources, broken-link outreach), and a "looks free, is spam" list.
AI search. What Google says is required (nothing extra), how other engines choose sources, and how to measure visibility as a percentage of runs, never a "rank".
Ten site categories: local services, agency, SaaS, e-commerce, publisher, free tools, apps, marketplaces, hospitality, healthcare.
## Scripts
All in skills/mt7-seo-king/scripts/. They need Node 18+ and nothing else, and they fetch raw HTML the way a non-rendering crawler does.
| Script | What it answers |
| diagnose.mjs | What is wrong with this site, and what should be fixed first? Redirect chains and loops, links to errors, http/www/slash duplicates, soft 404s, sitemap URLs that are noindexed, redirected or blocked, broken canonicals, duplicate and thin content, hreflang errors, missing required JSON-LD properties, intermittent 5xx, mixed content and sitemap limits. Each finding has evidence, a source and a verify step; --psi adds PageSpeed data. No made-up score |
Every recommendation carries an evidence tag. [G] means Google's own documentation, [DATA] a study or measured test with its sample size and date, and [PROVEN] independent practitioners agreeing. [CLAIM], [WARNED] and [INFERENCE] are labelled as such, so you can tell a rule from a hunch. Where community advice contradicts Google, the skill follows Google and says so. Nothing paid is required: no API keys, and nothing to install beyond Node 18 for the scripts. If you use OpenSEO, the skill can also draw on its live keyword, SERP and backlink data, and asks before any call that spends credits.
## What it does
Audits and diagnosis. A phased workflow (diagnose → foundations → intent map → pages → local and reviews → links → AI search → measure). Every phase has a gate to pass before moving on.
Diagnose mode. diagnose.mjs crawls a site the way a polite, non-rendering crawler does. It ranks what it finds from critical to info, and gives each finding its per-URL evidence, its Google source and a way to verify it. On crawler-test.com, a site built with known SEO faults, four runs found 160 of its 164 test issues.
Traffic drops. A dated ledger of Google core, spam and policy updates, each entry citing a Google URL, plus Search Console's own logging errors. --around shows what overlapped a drop before anyone blames "the algorithm".
Verify before you report. Audit rules that stop the usual false positives: "your sitemap lists noindexed pages" (it gets checked first), "no schema" (on a JS-rendered page), "duplicate Organization" (a nested parent organization isn't one).
Pages that rank without becoming spam. Templates for service, industry, city, case-study, comparison and how-to-choose pages, each with a legitimacy gate. A programmatic-SEO gate covers tools, conversions and comparisons built from one template. Each page type also has its "becomes a doorway page when..." line.
Links at no cost. A vetted catalogue of free link and mention sources with their link attributes, outreach kits (lists, digital PR, journalist sources, broken-link outreach), and a "looks free, is spam" list.
AI search. What Google says is required (nothing extra), how other engines choose sources, and how to measure visibility as a percentage of runs, never a "rank".
Ten site categories: local services, agency, SaaS, e-commerce, publisher, free tools, apps, marketplaces, hospitality, healthcare.
## Scripts
All in skills/mt7-seo-king/scripts/. They need Node 18+ and nothing else, and they fetch raw HTML the way a non-rendering crawler does.
| Script | What it answers |
| diagnose.mjs | What is wrong with this site, and what should be fixed first? Redirect chains and loops, links to errors, http/www/slash duplicates, soft 404s, sitemap URLs that are noindexed, redirected or blocked, broken canonicals, duplicate and thin content, hreflang errors, missing required JSON-LD properties, intermittent 5xx, mixed content and sitemap limits. Each finding has evidence, a source and a verify step; --psi adds PageSpeed data. No made-up score |
收录徽章(可选)
想把收录标记放进 README 就复制这行,可选,不影响收录。
[](https://anclum.com/zh/s/abduladnan-mt7-mt7-seo-king/)
- 来源
- github · https://github.com/abduladnan-mt7/Mt7-Seo-king
- 按领域浏览
- Agent 技能
- Traction
- 1 ★
- 收录来源
- 4 个独立来源
- 收录于
- 今天
同类推荐
所属领域与标签
每个标签都能点进去看同类的全部条目。数字是当前库里的真实条数。