The web, fetched through a rotating proxy pool.

VPNFail MCP gives AI agents a simple path to proxy-backed web access, returning clean Markdown, plain text, or original HTML.

Free anonymous access is live in public beta. The roadmap adds more capacity and control when the work needs it.

VPNFAIL MCP // STREAMABLE HTTP
$ vpnfail-mcp
ENDPOINT
https://mcp.vpn.fail/mcp

CLIENT CONFIGURATION
{
  "mcpServers": {
    "vpnfail": {
      "type": "http",
      "url": "https://mcp.vpn.fail/mcp"
    }
  }
}
Live exits719
TransportStreamable HTTP
Availability● public beta

Free launch access

Start with HTTP. Escalate when the page needs a browser.

Both retrieval modes are free during launch. Browser rendering receives one-fifth of the HTTP allowance because it uses substantially more compute.

01

Basic

Anonymous HTTP retrieval

  • 250 requests per hour
  • 1,000 requests per day
  • 5,000 requests per week
  • 10,000 requests per 30 days
  • Markdown, text, or HTML
  • Automatic quality metadata
  • Browser retry recommendations

● Public beta live · no account required

02

Browser

Rendered pages and verified geography

  • 50 requests per hour
  • 200 requests per day
  • 1,000 requests per week
  • 2,000 requests per 30 days
  • Chromium page rendering
  • Verified exit-country and city selection
  • Optional viewport screenshot

● Available now · future premium capability

03

Network

Direct proxy access and control

  • Authenticated SOCKS endpoints
  • Verified exit-country and city selection
  • Sticky session options
  • API access and usage controls
  • Team members and shared allowances
  • Priority support and capacity

For larger, location-aware workflows.

[✓] No account or payment is required during launch. Accounts and billing remain future work.

How it works

01Send a URL

Your agent calls scrape with a target page and output format.

02Rotate the route

VPNFail picks a live exit and retries across the pool when needed.

03Use the content

Receive a clean result ready for a workflow, model, or dataset.

Tools

scrape

Fetch one HTTP or HTTPS page as Markdown, text, or HTML.

usage

Inspect rolling anonymous quota consumption.

service_status

Inspect live exit availability and service health.