n8n

n8n’s HTTP Request node has a Proxy option, so any request a workflow makes can leave from a residential, datacenter or mobile IP in the country you pick:

• With the Proxmint node. It builds the proxy line for you, with targeting and sticky sessions, and also checks a connection and reads your balance, usage and prices. Best when the location or the number of IPs changes from run to run.
• Without it. Paste one proxy URL into the HTTP Request node. Nothing to install, and enough when every run uses the same location.

The node is n8n-nodes-proxmint, and it has passed n8n’s review as a verified community node. Where it shows up in the nodes panel, search for Proxmint and install it from there; that works on n8n Cloud too. On self-hosted n8n you can also open Settings > Community nodes > Install and enter the package name.

Then create an API key under Settings in the dashboard. Keys start with pmk_ and can only read: the node cannot buy, top up or change targeting. In n8n, add a Proxmint API credential and paste the key.

The source is on GitHub and the package on npm. Each operation is one call to the customer API, which allows 60 requests a minute per key.

1. Add a Proxmint node: Proxy > Generate. Pick the product, set Protocol to HTTP and leave Format on URL.
2. Under Location, add a country and, if you need it, a state, city, ZIP or ASN. Leave it empty to use the targeting saved in the dashboard.
3. Connect an HTTP Request node after it. Under Options > Add option > Proxy, enter {{ $json.proxy }}.

Rotating gives one line, and every request through it gets a new IP. Sticky gives Count lines, each keeping its own IP for Sticky Minutes (up to 120). The HTTP Request node runs once per item, so a sticky batch spreads the requests over that many IPs, one per line.

Use HTTP, not SOCKS5. The HTTP Request node only takes proxy URLs that start with http://. Given a socks5:// line, or a line with no http:// in front, n8n skips the proxy and sends the request from the n8n server’s own IP, with no error in the workflow. HTTP lines still open HTTPS sites. To confirm the proxy is in use, send one request to https://api.ipify.org and compare the IP with your server’s.

Copy the username and password from your product page in the dashboard and enter this in the HTTP Request node’s Proxy option:

http://PROXY_USER:PROXY_PASS@gw.proxmint.com:823

Location and sessions go in the username, after two underscores (see targeting & sessions). n8n passes the ; between parameters to the gateway as it is, and the field takes expressions, so a session ID can come from the execution:

# A US IP in New York, new IP on every request
http://PROXY_USER__cr.us;city.newyork:PROXY_PASS@gw.proxmint.com:823

# One IP for the whole execution, for up to 30 minutes
http://PROXY_USER__cr.us;sessid.{{ $execution.id }};sessttl.30:PROXY_PASS@gw.proxmint.com:823

Put the password in the field only on a workflow you don’t share: anyone who can open the workflow can read it. The Proxmint node keeps it out of the workflow, in a credential.

Every operation only reads. Connection > Check sends one request of a few kilobytes through your product. Proxy > Generate and Account > Get return the proxy password, because a working line needs it, so keep the execution data private.

The node is marked usable as a tool, so an n8n AI Agent can call it to check a balance or build proxy lines.

• State, city, ZIP and ASN targeting can meter at twice the bandwidth; the targeting page says which products.
• The HTTP Request node starts every item’s request at once. Each product allows up to 2,000 concurrent connections, so for long item lists set Options > Batching to send them in smaller groups.
• Pages load over the destination ports the network forwards (see port access), and a short list of sites is refused outright (see blocked websites). When the gateway refuses a request, the error codes page says why.
• To route all of a self-hosted n8n’s outbound traffic through one proxy, n8n also reads the HTTP_PROXY, HTTPS_PROXY and NO_PROXY environment variables. That includes its calls to your other services, so the per-node option is usually the better fit.