Why you do not get banned on day one. The technical version.


X uses multi-layer bot protection. A simple "User-Agent + proxy" setup is detected in minutes. We implement 14 layers of masking.


<aside> 🌐 Network Layer

X uses TLS fingerprinting and header analysis. These 6 layers mask our traffic as Chrome.

Layer What it does
TLS Fingerprint: JA3/JA4 Chrome 145 impersonation through curl_cffi — the TLS handshake looks like Chrome
Header order Headers are sent in Chrome order through OrderedDict, not randomly
Client Hints sec-ch-ua, sec-ch-ua-platform, sec-ch-ua-mobile — as in Chrome
Accept-Language Adjusted to the proxy geolocation
Timezone Matches the proxy country
TLS extension order Random order of extensions in the TLS handshake

TLS Fingerprint: JA3/JA4

Header order

</aside>


<aside> 🧩 Application Layer

X checks cookies, tokens, and request signatures. These 4 layers imitate web-client behavior.

Layer What it does
x-client-transaction ID Generated like in a real X web client
Session warming On startup, the worker "opens" the home page and receives session cookies
ct0 refresh Refreshed every 30 minutes, like a real browser does
Cookie jar Full set of cookies, not only the required ones

Session warming

</aside>


<aside> 🎭 Behavioral Layer

X analyzes behavior patterns: speed, rhythm, and schedule. These 4 layers imitate a real person.

Layer What it does
Typing speed 50-90 WPM with jitter — a reply is "typed", not instantly appeared
Delays Log-normal distribution, not Gaussian; real people are not normally distributed
Diversity check 60-minute cooldown for phrases with similarity >70%
Sleep mode 7 hours of silence per day, like a real person

Log-normal Delays

</aside>