Proxy Authentication: Username Password Or IP Whitelist
A proxy that anyone can use is a proxy you’re paying for someone else to abuse. So every paid proxy has some way of deciding who’s allowed through, and almost every provider gives you the same two choices: a username and password, or a list of your own IP addresses that are allowed to connect without one.
Most operators pick whichever the setup guide shows first and never think about it again. That’s a pity, because the choice decides what breaks when your internet connection changes, who else can use your lines, what ends up written into files you share, and what record ties the proxy account back to you. None of it is exotic. It’s plumbing. But plumbing is where fleets spring leaks.
What authentication actually does
Authentication sits between you and the proxy server. It answers one question for the provider: should this connection be allowed to use this proxy, and whose account should it count against.
It doesn’t touch the relationship between the proxy and the website you’re visiting. The target site never sees your proxy password, and it never sees your whitelisted home address. It sees the proxy’s exit IP and whatever your browser sends through it. That’s worth saying clearly, because people sometimes worry that the authentication method changes how they look to a platform. It doesn’t, as long as the proxy works.
Where it matters is everything around that. How reliably the proxy works. Who else could use it. What gets exposed on the path between you and the provider. What’s left behind in configuration files and logs. And what the provider learns about you.
The two methods make different trade-offs on each of those, and neither is better everywhere. A solo operator on a fixed connection and a team of five working from laptops need different answers. So it helps to take each one apart.
How username and password auth works
With username and password authentication, every connection to the proxy carries credentials. The proxy checks them and either lets the connection through or refuses it.
For an HTTP proxy, the credentials usually travel in a header called Proxy-Authorization, using the Basic scheme. Basic is exactly what it sounds like. The username and password are joined with a colon and encoded in base64, which is a way of representing text, not a way of hiding it. Anyone who can read that header can decode it instantly.
For HTTPS websites, your browser first sends a CONNECT request to the proxy asking it to open a tunnel to the site, and the credentials ride on that CONNECT request. After that, the traffic inside the tunnel is encrypted between your browser and the website. But the CONNECT request itself, credentials included, goes to the proxy in plain text unless the connection to the proxy is also encrypted.
SOCKS5 has its own username and password method, defined in a short standard of its own. It’s also plain text on the wire.
So the credentials are only as private as the path between your machine and the proxy server. On your home connection, that path runs through your router, your ISP and whatever networks sit between them and the provider. On public Wi-Fi, it includes everyone else on that network. A few providers offer proxy endpoints you connect to over TLS, which encrypts that first hop, and that’s worth using where it’s available.
The upside is portability. The credentials work from anywhere. Move to a new connection and nothing changes.
How IP whitelisting works
With whitelisting, you tell the provider which of your IP addresses are allowed to use the proxy, and connections from those addresses go through with no credentials at all. Connections from anywhere else get refused.
It’s simpler at the moment of use. No password to type, no header to carry, nothing for a tool to support. Any software that can point at a host and port can use the proxy. That makes whitelisting the easy answer for tools with poor or no proxy authentication support, and there are more of those than you’d think.
It moves the security question from “who knows the password” to “who is behind that address”. And that’s the part people underthink. A whitelist doesn’t authenticate you. It authenticates a network location. Anyone who can send traffic from that location gets the same access you do.
On a dedicated server with its own static address, that’s fine. The address belongs to you and only your software sends from it. At home, it means everyone on your home network. In an office, everyone behind the office’s shared address. And on some residential connections, as we’ll get to shortly, it can mean strangers.
Whitelisting also has a practical ceiling. Providers usually cap the number of addresses you can list, and managing that list becomes its own chore when you have many machines on many connections.
The day your home IP changes
The most common whitelisting failure is a Tuesday morning when nothing works.
Most home broadband connections don’t have a fixed public address. The ISP assigns one, and it can change when the router restarts, when the lease expires, after maintenance, or for no reason you’ll ever be told. Some connections keep the same address for months. Others change weekly.
When it changes, the whitelist no longer matches. Every tool pointed at the proxy starts failing at once. Depending on the provider, you’ll see refused connections, timeouts or an error page from the proxy. Now you’re logging into the provider’s dashboard, finding your new address and adding it, while some of your sessions were half finished.
The bigger risk is how your tools behave in that gap. Some fail cleanly and just stop. Others fall back to connecting directly without the proxy, and then your accounts are suddenly visiting platforms from your real home address. That one silent fallback can do more damage than any password leak, because it hands the platform the one address you most wanted to keep out of the fleet.
If you whitelist a home connection, find out how often its address changes, and check how every tool in your stack behaves when the proxy refuses a connection. The right behaviour is to stop. Anything that quietly carries on is a problem to fix today.
Carrier grade NAT makes whitelisting risky
There’s a less obvious problem with whitelisting on some connections, and it’s worth understanding before you whitelist an address you don’t fully own.
Many ISPs, and nearly all mobile carriers, use carrier grade NAT. Instead of giving each customer their own public address, they put many customers behind one shared public address and translate between them. From the outside, all of those customers look like they come from the same IP.
Now imagine whitelisting that IP with your proxy provider. The provider sees connections from that address and lets them through. It has no way to tell your connections from those of the other customers sharing it. Anyone behind the same carrier address who happens to point at your proxy’s host and port gets through on your account.
In practice, a random neighbour on the same carrier is unlikely to stumble onto your proxy endpoint. But “unlikely to be found” is a weak protection for something billed by traffic or tied to your accounts’ reputation. And the shared address changes often on these networks, so the whitelist breaks frequently too.
You can often tell whether you’re behind carrier grade NAT by comparing the WAN address your router reports with the public address a lookup site shows you. If they differ, especially if the router’s address falls in the 100.64.0.0/10 range, you’re sharing.
The rule is simple. Only whitelist addresses that belong to you alone. For everything else, use credentials.
Credentials that carry configuration
Many providers do something clever with the username field. Instead of a plain account name, the username carries settings: which country to exit from, whether to keep a sticky session and for how long, which session ID to use, sometimes which city or carrier. The proxy reads those parameters from the username and routes accordingly.
It’s convenient. One endpoint, one password, and you get different behaviour by changing the username string. For fleets, it means each browser profile can carry its own session parameter and keep its own exit.
It also changes what a credential leak exposes. A leaked password on its own gives someone access to your traffic allowance. A leaked username-and-password string with session parameters tells them which sessions exist, how you route your accounts and, in some setups, lets them attach to the same sticky session as one of your profiles. Two different people using the same sticky session means two different sets of behaviour through the same exit at the same time, which is exactly the kind of overlap you’re paying to avoid.
Treat the full proxy string as a secret, parameters included. When you share a setup with a teammate, share it through whatever system you use for passwords, and give them their own session identifiers rather than copying yours. When you post a question on a forum about a proxy problem, blank out the whole string, not just the password.
How credentials leak in real fleets
Credential leaks in fleets rarely involve anyone breaking in. They’re almost always the result of credentials being copied into places nobody thinks of as sensitive.
The first is exported browser profiles. Antidetect browsers let you export a profile to move it or back it up, and the export usually includes the proxy settings, credentials and all. That file ends up in a shared drive, a chat message or a backup folder, and now the proxy password lives wherever that file lives.
The second is spreadsheets. A tab listing every account, its profile and its proxy string is useful. It’s also a complete map of your fleet and its access, and spreadsheets get shared with a click.
The third is screenshots. Someone sends a screenshot of a proxy settings screen to ask why it isn’t connecting, and the password is right there in the field.
Then there are scripts and repositories, where a proxy string gets hard-coded into automation and pushed somewhere with more readers than intended. And logs, where some tools print the full proxy URL, credentials included, when a connection fails.
The fix is mostly habit. Keep proxy strings in one place, a password manager or a secrets store, and reference them from there. Strip credentials from anything you export before it’s shared. Check what your tools log. And assume any credential that has been pasted into a chat is already public.
Sub-users and per-person credentials
The worst credential setup for a team is one proxy account with one password that everyone uses. It’s also the most common.
Many providers let you create sub-users under your main account. Each sub-user gets its own credentials, and often its own traffic limits, allowed IPs or product access. If your provider offers this, use it. Give each person their own credentials, and ideally each project or client its own set too.
This does three useful things. First, when someone leaves, you disable their credentials and nothing else changes. With a single shared password, offboarding means rotating it and updating every profile and script that uses it, which in practice means it doesn’t get done. Second, when traffic spikes or something misbehaves, the provider’s usage breakdown tells you whose credentials did it. Third, a leak is contained. A password posted in a screenshot compromises one person’s access, not the whole account.
If your provider doesn’t offer sub-users, the fallback is separate provider accounts for separate teams or clients. That’s more to manage, but it keeps a single mistake from exposing everything.
The same principle works for whitelisting. If each person works from their own static address, each address can be listed separately and removed on its own. What doesn’t work is a whitelist full of addresses nobody remembers adding.
Which tools support which method
The choice is often made for you by the tools, so it’s worth knowing their limits before you commit to a method.
Antidetect browsers generally support username and password authentication per profile, for both HTTP and SOCKS5, and handle the authentication prompt behind the scenes. This is the easy case. Each profile gets its own proxy string and nobody types anything.
Regular browsers are less friendly. Chromium-based browsers can be pointed at a proxy from the command line, but there’s no flag for credentials, so they fall back to a pop-up asking for the username and password, or you need an extension to supply them. Firefox handles proxy settings in its own preferences and also prompts for credentials.
Automation frameworks usually accept proxy credentials in their launch options. Some command-line tools accept credentials embedded in the proxy URL, which works but means they appear in your shell history and process list.
Android is the awkward one. The proxy settings in a phone’s Wi-Fi configuration take a host and port, with no field for credentials. Apps that respect the system proxy will fail to authenticate unless the proxy is whitelisted or a separate app handles the connection. So phones often end up on whitelisting, or on a proxy client app, or on a network where the proxy is applied upstream and the phone doesn’t need to know about it.
Before choosing, list every tool that will use the proxy and check which methods each one handles properly.
Reading the error when auth fails
When a proxy stops working, the first job is telling an authentication failure apart from everything else that can go wrong. The symptoms overlap, and people waste hours changing exits when the real problem is a password.
With credentials, a wrong or expired password on an HTTP proxy usually produces a specific response: status 407, Proxy Authentication Required. Browsers show it as a login pop-up that keeps coming back, or as an error page naming the proxy. Automation tools report it as a 407 in their logs. If you see 407, the exit is irrelevant. The proxy never got as far as choosing one.
With whitelisting, failures look vaguer. Some providers refuse the connection outright, some return a 403 or a custom error page, and some just let the connection hang until it times out. A timeout is the confusing case, because it looks exactly like a dead exit or an overloaded server.
A quick way to separate them is to test from the command line with a simple request through the proxy to an IP lookup page, printing the response headers. A 407 or an explicit provider error points at authentication. A successful connection that shows an unexpected exit points at routing or session parameters. A timeout from a whitelisted machine is worth checking against your current public address before anything else, because a changed home IP is the most common cause.
SOCKS5 failures are terser. The protocol returns a short failure code during the handshake, and most tools translate that into a generic “connection failed” message. Check the tool’s verbose or debug output, which usually shows whether the failure happened at the authentication step.
Keep a note of what a healthy connection looks like for each provider you use. When something breaks, comparing against a known-good result is faster than guessing.
The IPv6 gap in whitelists
A whitelist that looks right can still fail because your machine isn’t connecting from the address you think it is.
Many home connections are dual stack now, with both an IPv4 and an IPv6 address. When your machine connects to the provider’s proxy hostname, it may use either, depending on what the hostname resolves to and how your operating system prefers to connect. If the provider’s endpoint has an IPv6 address and your machine uses it, the connection arrives from your IPv6 address, which isn’t on the whitelist if you only added the IPv4 one.
The result is a whitelist that works on one machine and not another on the same network, or that works some of the time. It’s maddening to debug if you don’t know to look for it.
The fix depends on the provider. Some let you whitelist IPv6 addresses or ranges. Others only accept IPv4, in which case you can connect to the endpoint’s IPv4 address directly, or configure the tool to prefer IPv4. Check what a lookup page reports as your address from the same machine, and check both address families if your connection has both.
There’s a related privacy point. IPv6 addresses on home connections often stay stable for long periods and can be closer to identifying a specific household than a shared IPv4 address. Adding one to a provider dashboard is a more precise record than adding the IPv4 address in front of it. That’s a small thing, but it’s worth knowing which address you’re handing over.
Local forwarders that hold the credentials
When a tool can’t handle proxy credentials, a common workaround is a local forwarder. You run a small proxy program on your own machine, point the tool at it without any authentication, and the forwarder connects onward to the real proxy with the username and password.
It’s a good pattern. The credentials live in one place, the forwarder’s configuration, instead of being copied into every tool. It lets tools that only understand host and port use an authenticated proxy. And it gives you one spot to change when you rotate a password.
It also creates a new risk that catches people out. The local forwarder is itself an unauthenticated proxy. If it listens only on your own machine, on 127.0.0.1, that’s fine, because only software on that machine can reach it. If it listens on all interfaces, which some tools do by default, then every device on your network can use it, and so can anyone who reaches your network. On a laptop at a café, that’s everyone on the café Wi-Fi, all using your paid proxy with your credentials attached.
When you set up a forwarder, check what address it binds to and set it explicitly to localhost unless you have a reason not to. If other machines genuinely need it, restrict access with a firewall rule rather than leaving it open.
The same care applies to forwarders running on a server. An unauthenticated proxy on a public server gets found by scanners quickly, and then it’s carrying traffic you’ll be blamed for.
Mixing methods without making a mess
Many providers let you use both methods on the same account. Whitelisted addresses get through without credentials, and anything else can still authenticate with a username and password. That’s useful, and it’s also a way to lose track of who has access.
A sensible mix looks like this. Servers with static addresses that you control run automation through a whitelist, because there’s nothing to store and nothing to leak. Laptops, travelling team members and anything on a home or mobile connection use per-person credentials. Phones use whatever works for them, ideally a setup where the proxy is applied upstream so the device carries no credentials at all.
Write down which method each machine uses and why. When something breaks, the first question is which path the connection took, and you want to answer it from a record rather than from memory.
Review the whitelist regularly. Remove addresses you can’t account for. An old office address, a VPS you shut down months ago, the hotel connection someone added during a trip. Each stale entry is a door you forgot about, and on a shared or recycled address it may now belong to someone else entirely.
If the provider’s dashboard shows which addresses and credentials are actually in use, compare that with your records. Access that exists but isn’t being used is usually access that should be removed.
What the provider learns about you
Both methods tell the provider something, and it’s worth being deliberate about what.
With whitelisting, you hand the provider your own IP addresses directly. If you whitelist your home connection, the provider now holds a record connecting your proxy account to your home address. That’s not a threat in itself. Your provider needs to know who its customers are. But it’s a record, and records get breached, subpoenaed and sold when companies change hands. If keeping your own identity separate from the fleet matters to you, a dashboard entry listing your home IP is one more join.
With credentials, the provider sees your source address on every connection anyway, because that’s how networking works. The difference is that your address shows up as traffic metadata, which the provider may or may not keep, rather than as a setting on your account.
Either way, the provider sees which exits you use and how much traffic goes through them. Choose providers you’d trust with that information, and read what they say about logging.
For most operators this is a small consideration next to reliability and leaks. But if you run the fleet through a company, or you care about keeping personal and working infrastructure apart, route your own access through something that isn’t your home connection and authenticate with credentials.
Rotating credentials and offboarding
Credentials that never change eventually leak. Not through anything dramatic, just through the slow accumulation of copies: old exports, old screenshots, an old laptop, a former contractor’s notes.
Rotate proxy passwords on a schedule, and immediately after anyone with access leaves or any credential shows up somewhere it shouldn’t. With per-person sub-users, rotation is easy because each change affects one person’s setup. With a single shared password, it’s painful, which is the strongest argument for sub-users in the first place.
Plan the rotation so it doesn’t break the fleet. Update the password in your secrets store, push the change to the profiles and scripts that use it, and only then disable the old password. Some providers let two passwords overlap briefly, which makes this painless. Where they don’t, do it at a quiet hour and check that every profile reconnects.
Offboarding is the same process with less room for delay. When someone stops working on the fleet, disable their credentials and remove any addresses they added to the whitelist the same day. Our piece on offboarding people who touch accounts goes further on the human side. For proxies, the test is simple. After someone leaves, could they still route traffic through your account from their own machine? If the answer is yes, you haven’t finished.
A Monday checklist
Here’s what I’d do this week if I hadn’t looked at proxy authentication in a while.
Open the provider dashboard and list every whitelisted address and every set of credentials. For each one, write down which machine or person uses it. Remove anything you can’t identify.
Check whether any whitelisted address is a home or mobile connection. If it is, find out whether it’s behind carrier grade NAT and how often it changes. Consider moving that machine to credentials.
Pick one tool that uses the proxy and deliberately break its access, by pausing the whitelist entry or changing its password. Watch what it does. If it connects directly instead of stopping, fix that before anything else.
Search your shared drive, chat history and repositories for your proxy hostname. Every hit that includes credentials is a leak to clean up.
If everyone shares one password, set up sub-users and move people across one at a time.
Set a date for the next rotation and put it in the calendar.
None of these steps takes long, and each one closes something that’s probably open right now.
The honest limit
Authentication protects your proxy account. It doesn’t make the proxy good, and it doesn’t make your accounts look legitimate. A perfectly secured proxy with a burned exit IP still gets flagged, and a sloppy profile behind a well-managed whitelist still links accounts. Choose the exit for the platform and the account first, then secure the access to it.
Within its own job, the trade-off is fairly clear. Whitelisting is simpler and leaves nothing to leak, but it’s only as safe as the address it trusts and it breaks when that address changes. Credentials are portable and per-person, but they need to be stored and shared carefully, and they travel in plain text unless the first hop is encrypted. Most fleets end up using both, deliberately, with a record of which is which.
I run Singapore mobile proxy farms myself, and the access side of it gets as much attention as the modems, because a proxy that leaks or silently fails costs more than one that’s slow. For the rest of the proxy stack, from picking the right type for each platform to catching leaks, start on the Multi Account Ops home page.
Get new guides and videos first — join the Telegram channel.