[Python] How to simulate faster human-like typing with Python (like pyautogui) without being...

I’m working on an experiment where my professor is creating an anti-cheat website that detects typing patterns.

In my setup, I send a request from the browser using an extension when I press a command (Ctrl + Alt + H). This triggers a Python script, which then injects text into the site at a defined position — specifically where the textbox is.

fetch("http://localhost:5000/type", {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify({
message: response, // IA response from Gemini API. But I get the text first: window.getSelection().toString()
cursor: cursor
})
})
.then(res => res.json())
.then(() => {
throwConsole(`Injected!`, 'ok');
})
.catch((e) => {
throwConsole(`Error: ${e}. Check the server from Python.`, 'error');
})



The injection works, but the site still detects the typing as artificial in some cases. I’m using pyautogui to simulate the typing, with a delay between characters, but it seems the detection is based on typing speed or event trust level.

I want to make the typing simulation faster but still human-like enough (randomized delays, pauses between words, maybe simulated typos) so that it’s less likely to be detected.

In python, I run a server to catch the information from JS.

import random
from flask import Flask, request, jsonify
from flask_cors import CORS
import pyautogui
import keyboard
import time

pyautogui.FAILSAFE = False

app = Flask(__name__)
CORS(app, supports_credentials=True)

screen_width, screen_height = pyautogui.size()

custom_x = int(screen_width / 2 + 100)
custom_y = int(screen_height / 2)

def simulate_typing(text):
for char in text:

# pyautogui.write(char, interval=0.6)
pyautogui.typewrite(char)
time.sleep(random.uniform(0.11, 0.12))



@app.route("/type", methods=["POST"])
def type_message():
if request.method == "OPTIONS":
response = app.make_response("")
response.headers["Access-Control-Allow-Origin"] = "*"
response.headers["Access-Control-Allow-Methods"] = "POST, OPTIONS"
response.headers["Access-Control-Allow-Headers"] = "Content-Type"
return response, 200

try:
data = request.get_json(force=True)
message = data.get("message", "")
cursor = data.get("cursor", {})

x = cursor.get("x", 0)
y = cursor.get("y", 0)

# Set the cursor in the textbox on Seia
x = custom_x if x == 0 else x
y = custom_y if y == 0 else y

pyautogui.click(x=x, y=y)
simulate_typing(message)


return jsonify({"status": "ok", "typed": message})
except Exception as e:
return jsonify({"status": "error", "message": str(e)}), 500


if __name__ == "__main__":
print("============================")
print(" AntiSeia is running! ")
print("============================")
app.run(host="0.0.0.0", port=5000)
time.sleep(3)



How can I adjust the Python code to improve the speed while keeping the typing natural and avoiding detection?

Here is the site of my professor: https://seia-eta.vercel.app/student

Code: QDS0E65I

Thanks in advance.

Continue reading...