Inspect what a model may receive
Prompt X-Ray is an ASCII Smuggling Detector and AI prompt security scanner for developers, security practitioners, and careful prompt authors. It helps reveal token estimates, invisible characters, bidirectional controls, mixed scripts, confusables, encoded content, and other anomalies in text before it is passed to a model.
Why hidden Unicode matters
Some Unicode characters are invisible or visually similar to ordinary text. Unicode Tags, zero-width characters, bidirectional controls, and confusable characters can make a prompt look different from the underlying string that software receives. Encoded or obfuscated content can also be difficult to review by eye.
Local-first inspection
The scanner is designed to process prompt text in the browser wherever technically possible. It does not need an AI API or prompt-processing backend. The tool presents local analysis as assistance for review, not as a final security verdict.
Know the limits
No static scanner can guarantee detection of every attack or security issue. Prompt X-Ray does not guarantee protection against prompt injection, ASCII smuggling, Unicode attacks, social engineering, or other threats. Treat results as one part of a broader secure-development and content-review process.