MCP 服务器

webpeel

The web data platform for AI agents. Fetch, search, crawl, extract, monitor, and screenshot any URL. 55+ domain extractors, 65-98% token savings. 7 MCP tools included.

README

The web data platform for AI agents — fetch, search, crawl, extract, monitor, screenshot, and research any URL.

Quick Start

npx webpeel "https://example.com"        # Clean markdown
npx webpeel search "AI trends 2025"       # Web search
npx webpeel crawl docs.example.com        # Crawl entire site

Get your free API key → · No credit card required · 500 requests/week free

Why WebPeel

65–98% token savings — domain-specific extractors strip boilerplate, ads, and nav before content reaches your agent
29 domain extractors — purpose-built parsers for Reddit, Wikipedia, GitHub, Hacker News, YouTube, ArXiv, Amazon, and 22 more
Zero-config Cloudflare bypass — 4-layer escalation stack handles TLS fingerprinting, edge proxying, and cache fallback automatically

Features

Feature	Command / API
Fetch any URL	`webpeel "url"`
Web search	`webpeel search "query"`
Crawl sites	`webpeel crawl "url" --max-pages 50`
Screenshots	`webpeel screenshot "url"`
Monitor changes	`webpeel monitor "url" --interval 300`
Browser actions	`--action 'click:.btn,wait:2000'`
YouTube transcripts	auto-detected
PDF extraction	auto-detected
MCP server	`webpeel mcp`
Schema extraction	`POST /v1/fetch` with `extract.schema`
Research agent	`POST /v1/agent`
Smart search	`POST /v1/search/smart`

MCP Integration

Give Claude, Cursor, or any MCP-compatible agent the ability to browse the web in one config change.

Claude Desktop (~/.claude/claude_desktop_config.json):

{
  "mcpServers": {
    "webpeel": {
      "command": "npx",
      "args": ["-y", "webpeel", "mcp"],
      "env": {
        "WEBPEEL_API_KEY": "wp_your_key_here"
      }
    }
  }
}

Cursor / VS Code (.cursor/mcp.json or .vscode/mcp.json):

{
  "mcpServers": {
    "webpeel": {
      "command": "npx",
      "args": ["-y", "webpeel", "mcp"],
      "env": {
        "WEBPEEL_API_KEY": "wp_your_key_here"
      }
    }
  }
}

Available MCP tools: webpeel_read, webpeel_find, webpeel_see, webpeel_extract, webpeel_monitor, webpeel_act, webpeel_crawl

Full MCP setup guide →

API Example

# Fetch any page — returns clean markdown + metadata
curl "https://api.webpeel.dev/v1/fetch?url=https://stripe.com/pricing" \
  -H "Authorization: Bearer $WEBPEEL_API_KEY"

{
  "url": "https://stripe.com/pricing",
  "markdown": "# Stripe Pricing\n\n**Integrated per-transaction fees**...",
  "metadata": {
    "title": "Pricing & Fees | Stripe",
    "tokens": 420,
    "tokensOriginal": 8200,
    "savingsPct": 94.9
  }
}

Full API reference →

Token Efficiency

WebPeel's 29 domain-specific extractors strip navigation, ads, sidebars, and boilerplate before sending content to your agent.

Site type	Raw HTML tokens	WebPeel tokens	Savings
News article	18,000	640	96%
Reddit thread	24,000	890	96%
Wikipedia page	31,000	2,100	93%
GitHub README	5,200	1,800	65%
E-commerce product	14,000	310	98%

Less context used = lower costs + faster inference + longer agent chains.

Security

WebPeel is built with security-first principles:

Helmet.js headers — HSTS, X-Frame-Options, nosniff, XSS protection on all responses
Webhook signing — HMAC-SHA256 signatures on all outbound webhooks
Audit logging — every API call logged with IP, key, and action
GDPR compliant — DELETE /v1/account for full data erasure
SSH hardened — Fail2Ban, MaxAuthTries, key-only auth on all infrastructure

Security policy →