Skip to content

Tomato

Hide encrypted data inside natural text

Visit official site
Category AI tools

What tomato is

The project is described as a steganography proof of concept built on minimum entropy coupling. It blends hidden information with LLM cover text so the result keeps natural language statistics.

What you can do with it

  • Encode a message into cover text
  • Decode a hidden message with the shared key
  • Set key length and prompt options

Key details

  • MEC merging minimizes joint entropy
  • LLM explanations assist the decode step
  • A command line tool handles both directions

Pros & cons

✓ What we like

  • Joint entropy is minimized to keep natural statistics
  • LLM context aware explanations aid decoding

! What to watch out for

  • Nvidia CUDA and drivers are required
  • Parameters such as the key must be set

FAQ

What code is used?

Minimum entropy coupling codes from ssokota.

How is data recovered?

MEC reverse decoupling extracts the hidden information.

Last reviewed: 2026-11-15

More AI office assistant tools

View all →

How we review