Hosted Tor · MCP · streamable-http
tor-mcp
Route HTTP requests through the Tor network from any MCP-compatible agent. Rotating anonymous exit IPs, .onion support, no request-body logging.
Honest scope: Tor exit nodes are blocked by Cloudflare on most big-brand sites (Indeed, Google, Amazon). Use tor-mcp for non-CF sites, .onion services, and IP-based rate-limit circumvention. Use browser-mcp when you actually need to hit Cloudflare-protected sites.
How it fits
Your agent → tor-mcp → Tor daemon → Tor network → Target site
(Claude, (JSON-RPC MCP) (SOCKS5) (~2M other (sees a
Cursor, Tor users) random exit
scripts) node IP)Limits (per API key)
Requests/day
100 max
Timeout
30s per request
Body cap
10 MB response
Cost
Free forever
Methods: GET / POST / HEAD only. No PUT / DELETE / PATCH (this skill is for anonymous fetching, not mutating resources). Higher quotas for aggregators — email shreyas.pavuluri@gmail.com.
Setup
- 1. Sign in with Google.
- 2. Generate an MCP API key on your dashboard.
- 3. Add this to Claude Desktop / Cursor / your agent:
{
"mcpServers": {
"tor": {
"url": "https://tor.regiq.in/api/mcp",
"headers": {
"Authorization": "Bearer <YOUR_KEY>"
}
}
}
}Tools
- •
tor_fetch(url, method?, headers?, body?)— HTTP request through Tor; returns status/headers/body - •
tor_get_exit_ip()— current exit node IP + country (via check.torproject.org) - •
tor_new_circuit()— force fresh Tor circuit; rate-limited by Tor to once per ~10s - •
tor_check(url)— HEAD probe; is target reachable through Tor?
Privacy
- • DNS resolves inside Tor (
socks5h://scheme) — no plaintext lookup from our host - • We log: target host, method, response status, duration, size. That's it.
- • We never log: full URLs, query strings, request bodies, response bodies, cookies, headers
- • Standard TorBrowser User-Agent sent by default — blend in with the fleet, not fingerprint yourself