saaspo-mcp
An MCP server that gives AI coding agents access to saaspo.com's SaaS web design inspiration library, enabling them to search pages and sections by style, industry, and type for grounded design decisions.
README
saaspo-mcp
An MCP server that gives AI coding agents access to saaspo.com — the SaaS web design inspiration library — plus a companion skill that makes the agent design landing pages like a senior designer: pull real references, deconstruct why they work, then mix and match.
3,000+ real SaaS pages · 700+ individual sections · tagged by style, industry, page type, section type and stack · with screenshots the agent can actually see.
Why
Coding agents produce visually average pages when they design from training data alone. Great designers steal like artists: they study world-class work, break down what makes it work, and recombine it. This server gives your agent that reference library, and the skill gives it the workflow — give the agent your copy, tell it the vibe, and it grounds every visual decision in real shipped designs instead of hallucinating.
Tools
| Tool | What it does |
|---|---|
get_design_vocabulary |
All searchable styles (Dark Mode, Bento, Neobrutalism…), industries, page types, section types, stacks — with example counts. Translates a "vibe" into filters. |
search_pages |
Search full pages by style/industry/page type/stack/free text. |
search_sections |
Search individual sections (Hero, Pricing, Testimonials, Footer…) for mix-and-match stealing. |
get_page_details |
Full record: live URL, tags, screenshot variants, the company's other pages and captured sections. |
get_screenshot |
Returns the actual screenshot as an image the agent can look at. Tall full-page captures are served in readable vertical segments. |
get_random_pages |
Random references for divergent exploration. |
Install
Register with Claude Code (no clone needed):
claude mcp add saaspo -- npx -y saaspo-mcp
Or in any MCP client config:
{
"mcpServers": {
"saaspo": {
"command": "npx",
"args": ["-y", "saaspo-mcp"]
}
}
}
To hack on it from source instead:
git clone https://github.com/doziben/Saaspo-MCP.git && cd Saaspo-MCP
npm install && npm run build
claude mcp add saaspo -- node /path/to/Saaspo-MCP/dist/index.js
The skill
skills/saaspo-designer/SKILL.md encodes the senior-designer workflow (brief → vocabulary → references → deconstruction → section plan → design tokens → implementation → self-critique). Install it for Claude Code:
mkdir -p ~/.claude/skills/saaspo-designer
curl -fsSL -o ~/.claude/skills/saaspo-designer/SKILL.md \
https://raw.githubusercontent.com/doziben/Saaspo-MCP/main/skills/saaspo-designer/SKILL.md
Then just give the agent your copy: "Design a landing page for my product with this copy — dark, technical, premium vibe." The skill triggers, pulls references through the MCP, and designs from them.
Dataset
The server ships with a pre-scraped index (data/). Saaspo adds new pages continuously; refresh whenever you like:
npm run scrape
Screenshots are never bundled — they're fetched on demand from saaspo's public CDN at view time.
How it works
Saaspo is a Webflow site. The scraper walks the public sitemap and parses each page's metadata (page type, styles, industries, stack, section tags, screenshot URLs) into a compact JSON index. At runtime the server searches the index in memory and fetches screenshots from the Webflow CDN, resizing and segmenting them (via sharp) so even 13,000px-tall full-page captures arrive as readable images.
Please be a good citizen: the scrape is a few thousand light HTML requests with modest concurrency. All content belongs to saaspo and the respective site owners — this tool is for design study, in the spirit of saaspo itself.
This project is not affiliated with or endorsed by saaspo. If you find the library useful, submit your sites and support them.
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。