If you discover a security vulnerability in Julius, please report it responsibly:

1. Do NOT open a public GitHub issue
2. Email security@praetorian.com with:
   • Description of the vulnerability
   • Steps to reproduce
   • Potential impact assessment
   • Any suggested fixes (optional)

We will respond within 48 hours and work with you to understand and address the issue.

Julius is an HTTP-based LLM service fingerprinting tool. It performs passive identification by:

• Sending standard HTTP GET/POST requests to target endpoints
• Analyzing HTTP response status codes, headers, and body content
• Matching response patterns against known LLM service signatures
• Reporting identified services with confidence levels

Julius is designed as a passive identification tool. It does NOT:

Julius generates HTTP traffic that resembles normal API client behavior:

• Standard HTTP methods (GET, POST)
• Common API paths (/api/tags, /health, /v1/models)
• JSON content types
• No malformed requests or fuzzing payloads
• Configurable concurrency and timeouts

Julius is intended exclusively for:

1. Authorized penetration testing - With written permission from system owners
2. Security assessments - As part of legitimate security audits
3. Asset inventory - On networks you own or administer
4. Research and education - In controlled lab environments

Always ensure you have explicit authorization before scanning any targets:

• Obtain written permission from the system owner
• Verify the scope of authorized testing
• Document authorization for your records
• Follow your organization's security testing policies

Unauthorized scanning of computer systems may violate:

• Computer Fraud and Abuse Act (CFAA) in the United States
• Computer Misuse Act in the United Kingdom
• Similar cybercrime laws in other jurisdictions

You are responsible for ensuring your use of Julius is lawful and authorized.

If Julius identifies exposed LLM services on your network, consider:

1. Verify authorization - Confirm the service should be network-accessible
2. Check authentication - Ensure API endpoints require authentication
3. Review network segmentation - Limit exposure to authorized networks
4. Audit access logs - Check for unauthorized access attempts

• Default bindings to 0.0.0.0 - Exposes service to all network interfaces
• Missing authentication - Allows unauthorized model access
• Exposed management APIs - Enables model loading/unloading by attackers
• Public cloud deployments - Internet-accessible LLM endpoints without auth

Julius uses these third-party dependencies:

All dependencies are regularly reviewed for security vulnerabilities.

Security updates will be released as patch versions (e.g., 0.1.1) and announced via:

• GitHub Releases
• GitHub Security Advisories (for critical issues)