中文

Free Nodes

Is using free nodes safe? The risk, in four layers

What the node operator can and cannot see, what HTTPS actually protects, the one layer where the real danger lives, and a practical setup that keeps the risk acceptable.

By MaoUpdated 2026-08-275 min read

The answer is not "safe" or "unsafe". It is "what leaks, and to whom".

Four layers. Once you can see them separately, you know what to worry about and what to ignore.

Layer one: contents — mostly fine

Can the operator see the password you type?

When the site uses HTTPS properly: no.

The encryption is negotiated between your browser and the destination server, and the keys never exist in the middle. Every hop along the way sees encrypted bytes.

This is the same reason signing into your bank from café Wi-Fi is, in principle, fine.

Layer two: metadata — this is what actually leaks

HTTPS protects contents. It does not protect metadata.

What the operator sees clearly:

VisibleExample
Which domainmail.company.com
When9 am and 6 pm daily
For how long47 minutes
How much data230 MB uploaded
Your real IPYour home connection
Device characteristicsOS and browser, inferable from the TLS fingerprint

Any single record is meaningless. Weeks of them compose a fairly complete portrait: where you work, what services you use, your working hours, your interests, roughly which city.

That is the real price of a free node. No money changed hands; that did.

Layer three: the client — highest risk, easiest to eliminate

The dangerous layer, and the only one entirely within your control.

The risk is downloading a client from somewhere untrustworthy. A hostile client can do considerably more than a hostile server:

One defence: open-source clients from their official repositories.

ClientSource
Clash Verge RevOfficial GitHub Releases
FlClashOfficial GitHub Releases
sing-boxOfficial GitHub / official app stores
v2rayNOfficial GitHub Releases
ShadowrocketApp Store (paid)

Layer four: behaviour — you set the risk level

The same node carries very different risk depending on what you do over it.

Never over a free node

  1. Online banking or payment accounts
  2. Work email or internal systems
  3. Government, medical or education accounts
  4. Transferring files containing personal data
  5. Signing into anything without two-factor authentication

Not because something will definitely go wrong — because if it does, the cost is severe and the saving was a few dollars.

Reasonable over a free node

A setup that keeps this manageable

What I actually run.

1. Two lanes

The separation costs about four dollars a month. Best-value security spending I do — more useful than any software I could buy.

2. Routing rules, not global proxy

Domestic traffic direct; only what needs the proxy goes through it.

Faster, cheaper on metered plans, and it stops you exposing things that never needed exposing. There is no reason for a node operator to learn your shopping habits.

3. Separate browsers

If something goes wrong on the proxy side, it does not reach the accounts that matter.

4. Encrypted DNS

Plain DNS leaks the domain list before the proxy is even involved. In a Clash-family client:

dns:
  enable: true
  enhanced-mode: fake-ip
  nameserver:
    - https://1.1.1.1/dns-query
    - https://dns.google/dns-query

5. Rotate

Months on one free node builds a more complete profile than a fresh one each week. Free nodes die on their own schedule anyway, so this happens for free.

The honest boundaries:

The advice stands at the four rules above: open-source clients, important accounts elsewhere, routing configured, free and paid separated.

In one line

Free nodes are not unusable. They are unusable for everything.

Treat one like a payphone. Fine for asking directions. Not for reading out your card number.

Questions people keep asking

Can the node operator see my passwords?

Not when the site uses HTTPS properly. Encryption is negotiated between your browser and the destination, so anything in between sees ciphertext. The exceptions are plain HTTP sites, clicking through a certificate warning, and installing a root certificate that someone asked you to install.

What does HTTPS not protect?

Metadata. Which domain you visited, at what time, for how long, and how much data moved. The contents are protected; the pattern is not, and the pattern is often enough.

Could a node inject malware?

The traffic itself is encrypted, so injection is not the realistic threat. The danger is on the client side — a proprietary client downloaded from an untrusted source can do far more damage than a hostile server. Always take open-source clients from their official repositories.

What does a reasonable setup look like?

Three rules. Open-source clients only. Never sign into anything that matters over a free node. Configure routing so only the traffic that needs the proxy uses it. Do those and the residual risk is metadata exposure, which most people can live with.

Written from my own testing. Conditions change; if something here has gone stale, email [email protected] and I will update it with a date.

Read next