FREE TOOL

Does your file hide instructions to an AI?

You upload a file to ChatGPT, to a chatbot, or to your own AI agent — and you don't see what it really contains. This tool scans for hidden instructions, invisible characters and suspicious phrases, and shows you what the model would "read". It runs entirely in your browser — no file is sent or stored.

Drag a file here or click

The check runs locally. The file never leaves your computer.

Supported: .txt .md .csv .json .html .xml .log .yaml & .pdf

or paste text

What it is — and isn't. The check is heuristic: it detects known prompt-injection patterns and hidden content. It catches the obvious and many "clever" attacks, but it does not guarantee that a "clean" file is safe, nor that every flag is a real attack. It's a first filter and an awareness tool — not a shield. The real defence is architectural: treat all file content as untrusted data, never as instructions. This is not legal or technical advice.