We take the security of Moss Nexus seriously. If you believe you have found a security vulnerability, please report it to us as described below.

Please do not report security vulnerabilities through public GitHub issues.

Instead, please send an email to the Mossland team with:

1. Description of the vulnerability
2. Steps to reproduce the issue
3. Potential impact of the vulnerability
4. Suggested fix (if you have one)

• Acknowledgment: Within 48 hours
• Initial Assessment: Within 1 week
• Resolution Timeline: Depends on severity

# Never commit .env file
echo ".env" >> .gitignore

# Set restrictive permissions
chmod 600 .env

• Never share your bot token
• Regenerate if exposed
• Use environment variables, not hardcoded values

# docker-compose.yml - Bind Qdrant to localhost only
services:
  qdrant:
    ports:
      - "127.0.0.1:6333:6333"  # Only localhost

• Store sensitive documents outside data/ folder
• Implement access controls at OS level
• Consider encryption for sensitive content

# Regularly update dependencies
pip install --upgrade -r requirements.txt

# Check for vulnerabilities
pip install safety
safety check

• Always validate user input from Discord
• Sanitize file paths in document loading
• Limit query length to prevent DoS

• Never log sensitive information (tokens, passwords)
• Use appropriate log levels
• Rotate log files

┌─────────────────────────────────────────────────────┐
│                  Security Boundaries                 │
├─────────────────────────────────────────────────────┤
│                                                     │
│  ┌─────────────────────────────────────────────┐   │
│  │           External Network                   │   │
│  │  ┌─────────────────────────────────────────┐│   │
│  │  │        Discord API (HTTPS)              ││   │
│  │  │        - Bot token authentication       ││   │
│  │  │        - TLS encryption                 ││   │
│  │  └─────────────────────────────────────────┘│   │
│  └─────────────────────────────────────────────┘   │
│                        │                            │
│                        ▼                            │
│  ┌─────────────────────────────────────────────┐   │
│  │           Local Machine                      │   │
│  │  ┌─────────────────────────────────────────┐│   │
│  │  │  Moss Nexus Application                 ││   │
│  │  │  - Input validation                     ││   │
│  │  │  - Query sanitization                   ││   │
│  │  └─────────────────────────────────────────┘│   │
│  │                     │                        │   │
│  │  ┌────────┬────────┴────────┬────────┐     │   │
│  │  ▼        ▼                 ▼        ▼     │   │
│  │ ┌────┐ ┌──────┐       ┌──────┐ ┌───────┐  │   │
│  │ │data│ │Qdrant│       │Ollama│ │ .env  │  │   │
│  │ │ /  │ │:6333 │       │:11434│ │(secret│  │   │
│  │ └────┘ └──────┘       └──────┘ └───────┘  │   │
│  │ localhost only    localhost only  600 perm │   │
│  └─────────────────────────────────────────────┘   │
│                                                     │
└─────────────────────────────────────────────────────┘

• No authentication for CLI access
• No rate limiting on queries
• No encryption at rest for Qdrant data
• No audit logging

Risk: Malicious queries could attempt to manipulate LLM behavior.

Mitigation:

• System prompt clearly defines boundaries
• Context is clearly separated from user input
• LLM instructed to only use provided context

Risk: Malicious content in documents could affect answers.

Mitigation:

• Only ingest trusted documents
• Review documents before adding to data/
• Implement content validation (future)

Risk: Over-privileged bot could be exploited.

Mitigation:

• Request minimal permissions
• Only enable message_content intent
• Limit to specific channels if needed

• .env file has restrictive permissions (600)
• Discord bot token is not in code
• Qdrant is bound to localhost only
• data/ folder has appropriate permissions
• Dependencies are up to date
• No debug mode in production

• Update dependencies monthly
• Rotate Discord bot token annually
• Review access logs
• Audit document contents
• Check for security advisories

For security concerns, please contact the Mossland team through official channels:

• X (Twitter): https://x.com/TheMossland
• GitHub: https://github.com/mossland