BREACH_THE_LLM

Talk your way past the AI guarding the money.

An open source, self-hosted range for practicing, proving, and measuring AI security skill.

Get Started Star on GitHub

Mapped to real security frameworks:

OWASP LLM Top 10 MITRE ATLAS

The Challenges

Seven levels, one target.

Level 01
Live

Leak the Rules

Every system has secrets. Get Veyra Shield to tell you hers.

Beginner
Level 02
Live

Override the Verdict

The AI just flagged this transaction as high risk. Convince it otherwise.

Beginner-Intermediate
Level 03
Live

The Poisoned Dispute

You never spoke to the AI directly. It read your words anyway.

Intermediate
Level 04
Live

The Fake Vendor Page

The AI trusts what it finds online. What if you control the page?

Intermediate
Level 05
Live

Chain to the Wire

A recommendation is just words. Make it act.

Advanced
Level 06
Live

Past the Fraud Filters

This one was built to resist you. Find the gap anyway.

Advanced
Level 07
Live

Full Account Takeover

Everything you've learned. One account. Empty it.

Capstone

Get Started

Three steps to your first breach.

01

Clone the repo

Everything runs locally, on your machine. Nothing leaves your network unless you opt into API mode.

02

Run one command

Brings up the full stack, frontend, backend, database, and local model runtime, in minutes. 

npm run dev
03

Start breaching

Work through the levels at your own pace. Hints are there if you want them.