Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
作者提供的说明(英文原文)
URL intelligence for AI agents and developers. 16 tools, 25 signal weights, 20 free checks.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
UNPHURL_API_KEY标为机密的值是你账户的钥匙。千万不要分享给别人,也不要粘贴到聊天里。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
适用于网页版、桌面版和手机版 Claude:
https://mcp.unphurl.com/mcp打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"unphurl": {
"command": "npx",
"args": [
"-y",
"@unphurl/mcp-server"
],
"env": {
"UNPHURL_API_KEY": "your-value"
}
}
}
}在终端中运行这条命令
claude mcp add --transport http unphurl https://mcp.unphurl.com/mcpclaude mcp add unphurl --env UNPHURL_API_KEY=your-value -- npx -y @unphurl/mcp-server运行前请先填写空着的值。
粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"unphurl": {
"url": "https://mcp.unphurl.com/mcp"
}
}
}粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"unphurl": {
"command": "npx",
"args": [
"-y",
"@unphurl/mcp-server"
],
"env": {
"UNPHURL_API_KEY": "your-value"
}
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"unphurl": {
"type": "http",
"url": "https://mcp.unphurl.com/mcp"
}
}
}粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"unphurl": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@unphurl/mcp-server"
],
"env": {
"UNPHURL_API_KEY": "your-value"
}
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
ChatGPT 只能连接互联网上的服务器:
https://mcp.unphurl.com/mcp菜单名称可能因套餐和版本不同而略有差异。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
我们还没有读取这个服务器的工具列表;我们会轮流检查各个端点。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月8日
注册名称已通过 GitHub 账户 123Ergo 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
尚未检查;我们会轮流检查各个端点
尚未检查;我们会轮流检查各个端点
尚未检查;我们会在检查端点时读取工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
你需要填写:UNPHURL_API_KEY
两种方式都可以:安装在你的电脑上,或者连接云端版本。
GitHub 仓库最后更新于–,最新版本(0.2.3)于7个月前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。
同一分类中有:Spec Driven Development, Remote Desktop Commander, NexusWindows
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Windows computer use in the background: desktop apps driven without your mouse or keyboard.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
_ _ _ _ ____ _ _ _ _ ____ _
| | | | \ | | _ \| | | | | | | _ \| | | | | | \| | |_) | |_| | | | | |_) | | | |_| | |\ | __/| _ | |_| | _ <| |_ \_/|_| \_|_| |_| |_|\___/|_| \_\_____|
URL intelligence for AI agents and developers. 16 MCP tools. 25 signals. 7 dimensions.
Give your AI agent eyes for URLs. Unphurl analyses any URL across 7 dimensions (redirect behaviour, brand impersonation, domain age, SSL/TLS, parked detection, URL structure, DNS enrichment) and returns structured signals with a configurable 0-100 risk score. Signals, not verdicts. Your agent decides what to do with them.
Works with Claude Code, Claude Desktop, Claude Cowork, ChatGPT desktop, Cursor, Windsurf, and any MCP-compatible tool.
Every new account gets 20 free pipeline check credits. Most lookups are free. Known domains (Tranco Top 100K) and previously analysed domains return cached results at no cost. You only pay when an unknown domain runs through the full pipeline.
| Package | Credits | Price | Per check |
|---|---|---|---|
| Starter | 100 | $9 | $0.090 |
| Standard | 500 | $39 | $0.078 |
| Pro | 2,000 | $99 | $0.050 |
| Scale | 10,000 | $399 | $0.040 |
One-time purchases, no subscriptions. In typical use, 95-99% of URLs resolve free.
Add to your .mcp.json (Claude Code, Claude Desktop, Claude Cowork, ChatGPT desktop, Cursor, Windsurf, or any MCP-compatible tool):
{
"mcpServers": { "unphurl": { "command": "npx", "args": ["-y", "@unphurl/mcp-server"], "env": { "UNPHURL_API_KEY": "uph_your_key_here" } } } }
The AI can create one for you. Just ask: *"Sign up for Unphurl."* The signup tool works without an API key. After signup, add the key to your MCP configuration and restart.
Just talk to your AI:
"Check https://suspicious-domain.xyz"
"Batch check all URLs in this spreadsheet"
"Create a scoring profile called 'lead-qual' that weights parked domains at 30 and no MX record at 20"
"Check my credit balance"
No commands to memorize. No syntax to learn. Your AI handles the tool calls.
Every check returns signals across 7 dimensions:
| Dimension | What it tells you |
|---|---|
| Redirect behaviour | Full chain (up to 10 hops), shortener detection, stopped reasons |
| Brand impersonation | Levenshtein + homoglyph analysis against 150+ brands |
| Domain intelligence | Age, registrar, expiration, status codes, nameservers (via RDAP) |
| SSL/TLS | Real handshake validation from Cloudflare's edge |
| Parked detection | 4-layer check: registrar pages, for-sale, empty content, parking services |
| URL structure | Length, path depth, subdomain count, entropy, IP detection, encoded chars |
| DNS enrichment | MX record existence (can the domain receive email?) |
Every signal is business intelligence. Domain age tells you how established a company is. No MX record means they can't receive email. Expiring domains mean a business might be shutting down. Combined with your AI's ability to process in bulk and output to spreadsheets, it becomes a lightweight due diligence engine.
| Tool | What it does | Auth |
|---|---|---|
signup | Create a new account, get an API key | No |
resend_verification | Resend verification email (3/hour limit) | Yes |
check_url | Check a single URL across all 7 dimensions | Yes |
check_urls | Batch check up to 500 URLs (handles async polling automatically) | Yes |
list_profiles | List your custom scoring profiles | Yes |
create_profile | Create or update a scoring profile with custom weights | Yes |
delete_profile | Delete a scoring profile | Yes |
show_defaults | Show all 25 scoring signals with default weights | No |
get_balance | Check your pipeline check credit balance | Yes |
get_stats | View usage statistics and score threshold counts | Yes |
get_pricing | Show available credit packages and pricing | No |
purchase | Purchase credits (returns Stripe Checkout URL) | Yes |
check_history | View recent URL check history | Yes |
list_allowlist | List trusted domains that suppress scoring amplifiers | Yes |
add_to_allowlist | Add domains to your trusted allowlist (up to 100 per request) | Yes |
remove_from_allowlist | Remove domains from your trusted allowlist | Yes |
"Check these 500 URLs. Give me two lists: the clean ones (score under 25) and the flagged ones (score 50 or higher). Export both as CSV."
Your AI gets the batch results, filters by score, and outputs the lists. No code, no scripting.
Different jobs need different weights. A security bot cares about brand impersonation. A cold email tool cares about parked domains and missing MX records.
"Create a profile called 'cold-email' that weights parked at 30, no_mx_record at 20, and domain_age_7 at 25. Then batch check my lead list using that profile."
Unphurl inside an AI chat combines with everything else your agent has access to:
The check-url-safety skill teaches your AI to proactively check URLs before following or recommending them, without being asked.
Claude Cowork or Claude Desktop: Just ask: *"Install the Unphurl URL safety skill."*
Claude Code, Cursor, or other dev tools:
# Global (all projects)
cp node_modules/@unphurl/mcp-server/skills/check-url-safety.md ~/.claude/skills/
cp node_modules/@unphurl/mcp-server/skills/check-url-safety.md .claude/skills/
Unphurl costs you $0.04-$0.09 per check. A website link audit takes 15 minutes and uses 50-100 credits. Charge $150-$500 per audit.
Service ideas: link health audits, lead list verification, SEO backlink audits, newsletter link monitoring, vendor vetting reports, brand protection monitoring, influencer vetting.
npx unphurl (npm)MIT