Security Policy

Security vulnerability reporting and handling policy for Strix

Strix is a security testing tool, and we take the security of the tool itself very seriously. This policy outlines how to report security vulnerabilities and how we handle them.

Reporting a Vulnerability

What to Report

Please report any security vulnerabilities in:

• Strix CLI - The main command-line tool
• Docker sandbox - Container isolation and security
• LLM integrations - API key handling and data leakage
• Dependencies - Vulnerable third-party packages
• Infrastructure - Strix Platform (app.strix.ai) vulnerabilities

How to Report

Do not report security vulnerabilities through public GitHub issues, discussions, or Discord. This could put users at risk.

To report a security vulnerability:

1. Email us privately

   • Send to: [e-mail]
   • Subject: “Security Vulnerability Report”
   • Include “CONFIDENTIAL” in subject line

2. Use GitHub Security Advisories

   • Go to: github.com/usestrix/strix/security/advisories
   • Click “Report a vulnerability”
   • Fill out the private disclosure form

3. For critical issues, encrypt your report

   • Use our PGP key (see below)
   • Ensures sensitive details remain confidential

What to Include

Please provide:

1. Description - Clear explanation of the vulnerability
2. Impact - What an attacker could do with this vulnerability
3. Affected versions - Which versions are vulnerable
4. Steps to reproduce - Detailed reproduction instructions
5. Proof of concept - Code, commands, or screenshots demonstrating the issue
6. Suggested fix - If you have ideas for remediation
7. Your contact information - So we can follow up with questions

Example Report Template

**Vulnerability Type:** [e.g., Command Injection, API Key Leakage]

**Severity:** [Critical/High/Medium/Low]

**Affected Component:** [e.g., strix/tools/terminal.py]

**Affected Versions:** [e.g., 0.8.0 - 0.8.2]

**Description:**
[Clear explanation of the vulnerability]

**Impact:**
[What could an attacker do? What data is at risk?]

**Steps to Reproduce:**
1. [Step 1]
2. [Step 2]
3. [Step 3]

**Proof of Concept:**
```bash
# Include reproduction code here

Suggested Fix: [If you have remediation ideas]

Additional Context: [Any other relevant information]

Contact: [Your name/email for follow-up]

## Our Commitment

### Response Timeline

We commit to:

- **Initial response:** Within 24 hours
- **Vulnerability assessment:** Within 3 business days
- **Status update:** Every 7 days until resolved
- **Fix release:** Based on severity (see below)

### Severity-Based Response

| Severity | Examples | Fix Timeline |
|----------|----------|-------------|
| **Critical** | Remote code execution, API key exposure | 1-3 days |
| **High** | Authentication bypass, data leakage | 7-14 days |
| **Medium** | Denial of service, information disclosure | 30 days |
| **Low** | Minor information leaks, configuration issues | Next release |

### Coordinated Disclosure

We follow coordinated disclosure:

1. **Day 0:** Report received
2. **Day 1:** Acknowledge receipt
3. **Day 3:** Validate and assess severity
4. **Day 7:** Develop and test fix
5. **Day 14-90:** Release patch (based on severity)
6. **After release:** Public disclosure with credit

We request you:
- Don't publicly disclose until we release a fix
- Give us reasonable time to patch (typically 90 days)
- Work with us if you need disclosure sooner

## Security Updates

### How We Notify Users

When we release security patches:

1. **GitHub Security Advisories** - Official CVE assignments
2. **Release notes** - Marked as security release
3. **Discord announcements** - Notify community
4. **Email notifications** - For Strix Platform users
5. **Social media** - Twitter/X announcements

### How to Stay Updated

**Subscribe to security notifications:**

1. **Watch Strix repository**
   - Go to [github.com/usestrix/strix](https://github.com/usestrix/strix)
   - Click "Watch" → "Custom" → Check "Security alerts"

2. **Join Discord**
   - [discord.gg/strix-ai](https://discord.gg/strix-ai)
   - Follow #announcements channel

3. **Follow on Twitter/X**
   - [@strix_ai](https://x.com/strix_ai)
   - Security updates posted immediately

### Updating Strix

Keep Strix updated to latest version:

```bash
# Check current version
strix --version

# Update via pip
pip install --upgrade strix-agent

# Update via pipx
pipx upgrade strix-agent

# Verify update
strix --version

# Pull latest Docker image
docker pull usestrix/strix:latest

Security Best Practices

For Users

Protect your LLM API keys:

# Don't commit keys to git
echo '.env' >> .gitignore
echo 'cli-config.json' >> .gitignore

# Use environment variables
export LLM_API_KEY="your-key"

# Don't share config files
chmod 600 ~/.strix/cli-config.json

Secure scan results:

# Results may contain sensitive data
chmod 700 strix_runs/

# Encrypt before sharing
tar -czf - strix_runs/latest/ | gpg -c > scan-results.tar.gz.gpg

# Securely delete
srm -r strix_runs/latest/  # or rm -P on macOS

Use least privilege:

# Don't run Strix as root
id
# uid=1000(user) - Good
# uid=0(root) - Bad

# Docker group is sufficient
groups | grep docker

Isolate testing:

# Test in isolated environments
# Use VMs, containers, or separate networks
# Don't test production from your daily driver

For Contributors

Secure development practices:

1. No secrets in code

   • Use environment variables
   • Never commit API keys or tokens
   • Use .gitignore properly

2. Input validation

   • Sanitize all user input
   • Validate file paths
   • Escape shell commands properly

3. Dependency security

   # Audit dependencies
   poetry audit
   pip-audit
   
   # Keep dependencies updated
   poetry update
   

4. Code review

   • All PRs require review
   • Security-sensitive changes need extra scrutiny
   • Run security checks before merging

5. Run security checks

   # Before submitting PRs
   make security-check
   
   # Or manually
   bandit -r strix/
   safety check
   

Known Security Considerations

LLM API Data Sharing

What data is sent to LLM providers:

When you use cloud LLM providers (OpenAI, Anthropic, Google), Strix sends:

• Target source code and configuration
• HTTP requests and responses
• Error messages and stack traces
• Testing instructions and findings

Mitigation options:

# Use local models for sensitive testing
export STRIX_LLM="ollama/llama3.3:70b"

# Or private cloud deployments
export STRIX_LLM="azure/gpt-5-private"
export LLM_API_BASE="https://private.openai.azure.com"

# Or AWS Bedrock
export STRIX_LLM="bedrock/anthropic.claude-v4"

Docker Container Isolation

Security measures:

• Containers run with limited privileges
• No access to host filesystem by default
• Network isolation from host
• Resource limits enforced

Considerations:

• Docker daemon has root privileges
• Container escape vulnerabilities could affect host
• Shared kernel with host system

Best practices:

# Don't run Docker as root
sudo usermod -aG docker $USER

# Use rootless Docker for extra isolation (optional)
export DOCKER_HOST=unix://$XDG_RUNTIME_DIR/docker.sock

# Keep Docker updated
docker version

API Key Handling

How Strix stores keys:

• Environment variables (cleared on exit)
• Config file: ~/.strix/cli-config.json (file permissions 600)
• Never in logs or reports
• Never sent to Strix servers

Protect your keys:

# Set restrictive permissions
chmod 600 ~/.strix/cli-config.json

# Rotate keys regularly
# Revoke keys if compromised

# Use separate keys for Strix
# Don't reuse production API keys

Security Features

Built-in Protections

Strix includes several security features:

1. Sandboxed execution - All testing runs in Docker containers
2. Input validation - Sanitizes user inputs and file paths
3. Rate limiting - Prevents runaway API usage
4. Timeout controls - Prevents infinite loops
5. Error handling - Fails safely without exposing internals

Security Defaults

# Default security settings
SECURE_DEFAULTS = {
    "sandbox_enabled": True,
    "verify_ssl": True,
    "max_file_size": "100MB",
    "timeout": 300,  # 5 minutes
    "rate_limit": True,
}

Vulnerability Disclosure History

We maintain transparency about security issues:

Published Advisories

See github.com/usestrix/strix/security/advisories for:

• CVE assignments
• Detailed vulnerability descriptions
• Affected versions
• Patches and workarounds
• Credit to reporters

Hall of Fame

We recognize security researchers who responsibly disclose vulnerabilities:

• See our Security Researchers Hall of Fame
• Contributors receive public credit (unless they prefer anonymity)
• Acknowledgment in release notes and security advisories

PGP Key

For encrypted communication:

-----BEGIN PGP PUBLIC KEY BLOCK-----

[PGP key would be here]

Fingerprint: [Key fingerprint]
-----END PGP PUBLIC KEY BLOCK-----

Download from: github.com/usestrix/strix/security/pgp-key.asc

Bug Bounty Program

We currently don’t have a formal bug bounty program. However:

• We greatly appreciate security reports
• Researchers receive public acknowledgment
• Critical vulnerabilities may be eligible for rewards at our discretion

Interested in a bug bounty program? Let us know: [e-mail]

Questions?

For questions about this security policy:

• Security issues: [e-mail]
• General questions: [e-mail]
• Discord: discord.gg/strix-ai

Changes to This Policy

This security policy may be updated periodically. We’ll notify the community of significant changes through:

• GitHub repository updates
• Discord announcements
• Documentation changes

Last updated: March 2026