Anti-Bot Links Vision Solver

A universal, intelligent Vision LLM solver designed to bypass Anti-Bot Links captchas. This type of captcha is commonly found across various crypto faucets (e.g., VieFaucet, DutchyCorp, FaucetCrypto, ClaimCoin) where users are required to click a sequence of text or math operations in a specific order.

By utilizing an OpenAI-compatible Vision endpoint (such as Gemma or GPT-4V), this package maps the target text sequence to their physical coordinates/indices in the image purely via HTTP, skipping heavy browser automation frameworks.

Why a Separate Project?

While initially conceptualized alongside the VieFaucet GuessCaptcha Solver, Anti-Bot Links are a universal protection mechanism used across hundreds of faucet platforms. This project extracts that logic into a highly reusable, platform-agnostic library.

Features

• Platform Agnostic: Works on any faucet utilizing image-based anti-bot sequential clicking.
• Vision-Driven Mapping: Understands complex math (3 + 4 -> 7), weird fonts, and rotated text.
• Pure HTTP Layer: Designed to be injected directly into requests or curl_cffi workflows.

Installation

git clone https://github.com/IndraYuda13/antibotlinks-solver.git
cd antibotlinks-solver
pip install requests

Usage Concept

from antibotlinks_solver import AntiBotLinksSolver

solver = AntiBotLinksSolver(
    router_key="sk-you...key",
    router_url="https://9router.indrayuda.my.id/v1/chat/completions",
    model="gemma"
)

# Example sequence required by the site
sequence = ["Lion", "Monkey", "Dog"]

# Pass the base64 image containing the scattered buttons
clicks = solver.solve(sequence, base64_image_data)
print(clicks)
# Output: [{"word": "Lion", "index": 2}, {"word": "Monkey", "index": 0}, {"word": "Dog", "index": 3}]

Disclaimer

This project is intended for educational purposes, security research, and personal automation. Usage of this script against terms of service is done at your own risk.