This article details the complete process of developing an automated script for claiming free assets on Fab using multiple AI models. The author started with Google Antigravity and VSCode’s Chrome DevTools MCP but encountered Cloudflare protection barriers. Subsequently, with Copilot’s help, GPT-5.2 Codex was used to write a DOM manipulation script, which faced element selection challenges. Next, Claude 4.5 Opus was tasked with writing a hybrid script based on network requests, but it still missed some assets. Finally, Gemini 3 Pro provided an optimized solution that successfully achieved automation. The script was converted into a Tampermonkey plugin, adding a visual UI, error logging, and internationalization support. The article also reflects on the limitations of AI-assisted development, emphasizing that the choice of technical approach without proper context can affect project maintainability. This practical case study demonstrates the real-world applications and challenges of AI tools in development, offering valuable insights for developers.
