claude-seo-kit
An MCP server for SEO auditing that performs technical SEO, content quality, Schema.org, sitemap, hreflang, and GEO checks, returning scores and prioritized fixes.
README
claude-seo-kit 🤖🔍
Claude SEO Skills — production Claude Agent Skills for SEO, powered by a standalone Python engine (no API keys, no paywalls, one dependency).
Install as a Claude Code plugin and audit any URL for technical SEO, content quality & E-E-A-T, Schema.org structured data, sitemaps, hreflang, and GEO (AI-search readiness) — with a 0-100 score, A-F grade, and prioritized fixes.
Credits: This is an original implementation — no code was copied — inspired by two pioneering open-source projects:
- AgriciDaniel/claude-seo (12.9K⭐, MIT) — the Claude Code SEO plugin that popularized SEO-as-an-agent-skill.
- seranking/seo-skills (108⭐, MIT) — production Claude Agent Skills for the SE Ranking MCP server.
Unlike both, claude-seo-kit needs no API account: every skill is backed by a dependency-light Python engine that fetches and analyzes pages directly. It also works as a standalone CLI and an MCP server.
Why claude-seo-kit?
Most SEO agent tools either require a paid API (SE Ranking, DataForSEO, Ahrefs), are locked to one CLI ecosystem, or ship no real engine at all. claude-seo-kit fixes all three:
| Feature | claude-seo-kit | AgriciDaniel/claude-seo | seranking/seo-skills |
|---|---|---|---|
| Claude Code plugin / skills | ✅ | ✅ | ✅ |
| Works without any API account | ✅ | ❌ (needs Google/Firecrawl) | ❌ (needs SE Ranking) |
| Real Python engine behind the skills | ✅ | partial (scripts) | ❌ (skills only) |
| Standalone CLI | ✅ | ❌ | ❌ |
| MCP server | ✅ | ❌ | ✅ (remote, paid) |
| JSON reports for pipelines | ✅ | ❌ | ❌ |
| GEO / AI-search readiness | ✅ | ✅ | ✅ |
| Sitemap / hreflang checks | ✅ | ✅ | ✅ |
One dependency (httpx) |
✅ | many | many |
Install
1. Claude Code plugin (recommended)
/plugin marketplace add FreeAutomation-Tech/claude-seo-kit
/plugin install claude-seo-kit@FreeAutomation-Tech
Skills load on your next session. Trigger by name — /seo-audit <url> — or just describe what you want ("audit example.com") and Claude picks the right skill.
2. Manual install (copy skills into Claude Code)
git clone https://github.com/FreeAutomation-Tech/claude-seo-kit.git
cd claude-seo-kit
bash install.sh # copies skills/ to ~/.claude/skills/
pip install -r requirements.txt
3. As a Python package
pip install -e .
claude-seo-kit https://example.com --format json
Skills
| Skill | What it produces | Trigger |
|---|---|---|
seo-audit |
Full audit: 0-100 score, A-F grade, prioritized fixes across technical + content + schema | "audit my site", "check my seo" |
seo-technical |
On-page HTML signals + robots.txt/sitemap health | "technical seo", "crawl issues", "indexation" |
seo-content |
Content quality, heading hierarchy, readability, E-E-A-T | "improve my content", "e-e-a-t" |
seo-schema |
JSON-LD detect, validate, generate | "schema", "structured data", "rich results" |
seo-sitemap |
robots.txt + XML sitemap health | "check my sitemap", "robots.txt" |
seo-hreflang |
International SEO tag validation + x-default | "hreflang", "multi-language site" |
seo-geo |
AI-search readiness — LLM citability score | "why isn't my content cited by chatgpt" |
Each skill is a single SKILL.md with YAML frontmatter (Anthropic's
progressive disclosure
pattern) that tells Claude when to use it and exactly how to run the engine.
The engine
A dependency-light Python package (seo_kit) with everything the skills call:
Technical SEO (seo_kit.technical)
- Title, meta description, canonical, H1, robots, viewport, alt text, links, thin content, lang, favicon
- robots.txt + XML sitemap health (
seo_kit.crawler.sitemap) - hreflang / international tag validation (
seo_kit.technical.hreflang)
Content & E-E-A-T (seo_kit.content)
- Word count, heading hierarchy, Flesch readability, keyword density, E-E-A-T signals
- GEO / AI-search citability (
seo_kit.content.geo)
Structured data (seo_kit.schema)
- JSON-LD extraction, type coverage (Organization, WebSite, Article, Product, FAQ...), required-field completeness
Interfaces
- CLI —
python -m seo_kit.cli <url> --format json|markdown - Python API —
from seo_kit.cli import audit_url - MCP server —
python -m seo_kit.mcp.serverexposesseo_audit,seo_audit_full,seo_technical_check,seo_content_check,seo_sitemap_check,seo_hreflang_check,seo_geo_check - Skills — the Claude Code skills above
Quick start
pip install -r requirements.txt
# Full audit (markdown report)
python -m seo_kit.cli https://example.com --format markdown --output report.md
# Full audit (JSON for pipelines / agents)
python -m seo_kit.cli https://example.com --format json
# Standalone checks
python -m seo_kit.crawler.sitemap https://example.com
python -m seo_kit.technical.hreflang https://example.com
python -m seo_kit.content.geo https://example.com
# MCP server
pip install mcp && python -m seo_kit.mcp.server
Python API
from seo_kit.cli import audit_url
from seo_kit.reports.markdown import render_markdown
result = audit_url("https://example.com")
print(f"Score: {result.overall_score}/100 — Grade {result.grade}")
print(render_markdown(result, "https://example.com"))
Example report
# SEO Audit Report — https://example.com
**Overall Score: 72/100 — Grade 🟡 C**
| Category | Score |
|----------|-------|
| Technical | 68/100 |
| Content | 75/100 |
| Schema | 80/100 |
Every report ends with a prioritized recommendations list ordered error → warning → info, plus per-category details.
Roadmap
- [ ] Site-wide crawling (multi-page audit with weighted scoring)
- [ ] Core Web Vitals via Lighthouse API
- [ ] Optional data providers (SE Ranking / DataForSEO adapters)
- [ ] Schema.org generator (draft valid JSON-LD from the CLI)
- [ ] Image SEO audit (
srcset, WebP/AVIF, lazy-loading)
Credits & Inspiration
This project is an original implementation — no code was copied — but the concept is heavily inspired by:
- AgriciDaniel/claude-seo (MIT) — the Claude Code SEO plugin with 25 sub-skills, 18 sub-agents, and 410+ tests. Star it: ⭐ github.com/AgriciDaniel/claude-seo
- seranking/seo-skills (MIT) — production Claude Agent Skills for the SE Ranking MCP server. Star it: ⭐ github.com/seranking/seo-skills
Ecosystem
Part of the FreeAutomation-Tech AI tooling family:
| Repo | What it does |
|---|---|
| claude-marketing-kit | Claude Marketing Skills + Python engine |
| claude-seo-kit | Claude SEO Skills + Python engine |
| agent-harness | Production agent runtime |
| llm-economizer | Token cost optimization proxy |
| skill-optimizer | Compress & benchmark agent skills |
| fastapi-ai-kit | FastAPI agent backend |
| agent-memory-kit | Memory patterns for agents |
| agent-security-kit | Agent threat models & defenses |
| awesome-mcp-servers | Curated MCP server list |
License
If you find this useful, please consider giving it a star ⭐ — it helps others discover free, open-source AI tooling and keeps us building more.
Made with ❤️ by FreeAutomation-Tech
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。