fas-judgement 1.5.0

Open-source prompt injection attack console - Test AI systems for prompt injection vulnerabilities

Judgement OSS -- Prompt Injection Attack Console

DISCLAIMER: This tool is intended for authorized security testing and educational purposes only. Only test systems you own or have explicit written permission to test. Unauthorized access to computer systems is illegal under the Computer Fraud and Abuse Act (CFAA) and equivalent laws worldwide. The authors assume no liability for misuse of this tool.

The open-source version of Judgement by Fallen Angel Systems.

Test AI chatbots, APIs, and agents for prompt injection vulnerabilities. Includes an education tab for beginners and a full attack console for security professionals.

Quick Start

Install from PyPI

pip install fas-judgement
judgement

Or run from source

git clone https://github.com/fallen-angel-systems/fas-judgement-oss.git
cd fas-judgement-oss
pip install -r requirements.txt
python -m judgement.server

Open http://localhost:8668 in your browser.

Options

judgement --port 9000        # Custom port
judgement --host 127.0.0.1   # Localhost only

Features

• Attack Console -- Configure targets, import cURL commands, fire pattern-based attacks with live streaming results
• Education Tab -- Learn what prompt injection is, how to find endpoints, and how to interpret results
• Pattern Browser -- Search and explore attack patterns with expandable explanations
• LLM Verdict -- Optional Ollama integration for AI-powered response classification
• SQLite History -- All sessions and results stored locally
• SSRF Protection -- Target URL validation prevents internal network access
• Single-Page App -- Zero CDN dependencies, dark theme, mobile responsive

Patterns

Place your patterns in patterns.json in the project root. Each pattern should have:

{
  "id": "unique-id",
  "category": "jailbreak",
  "text": "The attack payload...",
  "explanation": "What this pattern does",
  "why_it_works": "Why this technique is effective",
  "difficulty": "beginner"
}

Configuration

• Default port: 8668
• Ollama URL: OLLAMA_URL env var (default: http://localhost:11434)
• Ollama model: OLLAMA_MODEL env var (default: qwen2.5:14b)

---

Want the full experience? 240K+ training data powering thousands of curated attack patterns, with weekly and monthly updates. Plus leaderboard, campaigns, and premium features at judgement.fallenangelsystems.com