GreenAI Solutions 官网仓库:内置 MCP server 的零依赖静态站点
GreenAiSolution/greenai-solutions-group
GreenAI Solutions 将其亚利桑那州 AI 咨询与网页工作室的官网直接托管在该仓库,通过 GitHub Pages 提供服务,并内置一个 stdio MCP server,让 Claude 等 MCP 客户端能以工具调用方式查询站点结构、页面元数据、内容搜索、资源清单与内链检查。
The live website of GreenAI Solutions: an Arizona AI consulting and web studio, served straight from this repo via GitHub Pages — with a built-in MCP server so AI assistants can query the site like a database.
GreenAI Solutions (Gilbert, AZ) helps local businesses put AI to work — consulting, automation, social content, and custom web design. This repository is the production site: no framework, no build step, every page hand-built for speed.
What's in here
| Area | Files | Description |
|---|---|---|
| Marketing site | index.html, services.html, about.html, contact.html, testimonials.html |
Core public pages for greenaidigital.com |
| Legal & misc | privacy.html, terms.html, 404.html, thankyou.html |
Policies and utility pages |
| GreenAI Studio | studio.html + worker/ |
An in-browser AI tools suite backed by a Cloudflare Worker (keeps the Anthropic API key server-side; visitors use an access code) |
| Nexus AI portal | portal.html, portal-signin.html, portal-signup.html, portal-dashboard.html |
"AI employee network" portal concept (demo/prototype) |
| AETHER | aether/ |
Immersive 3D Three.js microsite for the autonomous-agents service (also maintained standalone at greenai-aether) |
| Showcases & demos | visuals*.html, experience.html, greenfit-ai-creator.html, content-studio.html, onboarding.html, marketing-campaign.html |
Interactive demos and internal marketing prototypes |
| MCP server | mcp-server/ |
Model Context Protocol server exposing the site to AI assistants |
The demo/prototype pages above are exactly that — working prototypes built to show what's possible, not claims of shipped client systems.
Features
- Zero-dependency static site — pure HTML/CSS/JS, fast on GitHub Pages, custom domain via
CNAME - Server-side AI proxy —
worker/is a Cloudflare Worker that fronts the Anthropic API for the Studio pages, deployed automatically by GitHub Actions (.github/workflows/deploy-worker.yml) - AI-native repo — ships its own MCP server so Claude (or any MCP client) can inspect pages, metadata, links, and assets with real tools instead of raw file reads
- SEO plumbing —
sitemap.xml,robots.txt, canonical/OG/Twitter meta on public pages
Architecture
graph LR
V[Visitor] --> GH[GitHub Pages<br/>greenaidigital.com]
GH -->|static HTML/CSS/JS| V
GH -->|studio.html fetch| CF[Cloudflare Worker<br/>worker/]
CF -->|API key stays server-side| ANT[Anthropic API]
A[AI assistant<br/>Claude, etc.] -->|stdio MCP| MCP[mcp-server/]
MCP -->|reads real files| REPO[(This repo)]
GA[GitHub Actions] -->|deploys on push to worker/| CF
Loading
Quickstart
git clone https://github.com/GreenAiSolution/greenai-solutions-group.git cd greenai-solutions-group python3 -m http.server 8080 # any static server works # open http://localhost:8080
The Studio's AI tools require the Cloudflare Worker — see worker/README.md for the 5-minute deploy.
MCP Server
This repo is AI-native: mcp-server/ contains a working stdio MCP server that turns the site into queryable tools.
| Tool | Purpose |
|---|---|
get_site_structure |
Every page with title, description, sitemap + robots status |
get_page_meta |
Full SEO/OG/Twitter metadata for one page |
search_content |
Text search across all site files with file:line results |
list_assets |
All assets with byte sizes, grouped by type |
check_internal_links |
Finds internal links pointing at missing files |
get_sitemap_report |
Cross-checks sitemap.xml against actual files |
cd mcp-server && npm install claude mcp add greenai-solutions-group -- node mcp-server/server.mjs
Details in mcp-server/README.md.
Tech stack
- HTML5 / CSS3 / vanilla JavaScript (no framework, no build step)
- Three.js (CDN) for the AETHER 3D microsite
- Cloudflare Workers + GitHub Actions for the Studio AI backend
- Node.js +
@modelcontextprotocol/sdkfor the MCP server - GitHub Pages hosting (master branch, custom domain)
Deployment
Pushing to master is a production deploy — GitHub Pages serves this branch at greenaidigital.com directly. Changes inside worker/ additionally trigger the Cloudflare Worker deploy workflow.
License
MIT © 2026 GreenAI Solutions
来源:GitHub · github.com