We release security updates for the following versions:

We take security seriously. If you discover a security vulnerability in XylonStudio, please report it responsibly.

Email: hello@xylonstud.io

DO NOT open a public GitHub issue for security vulnerabilities.

Please provide:

1. Description: What is the vulnerability?
2. Impact: What could an attacker do?
3. Steps to reproduce: How to trigger the vulnerability
4. Affected versions: Which versions are vulnerable
5. Suggested fix (optional): How might we fix it

Example:

Subject: SQL Injection in API endpoint

Description: The /api/design endpoint is vulnerable to SQL injection
Impact: Attacker could read/modify database
Steps: Send payload: {"spec": "'; DROP TABLE users--"}
Affected: v1.0.0 - v1.0.5
Suggested fix: Use parameterized queries

Critical vulnerabilities (RCE, data breach) are prioritized and fixed within 48-72 hours.

We follow coordinated disclosure:

1. You report the vulnerability privately
2. We acknowledge within 48 hours
3. We develop a fix
4. We release the patch
5. We publicly disclose (crediting you, if desired)

Typical timeline: 30-90 days from report to public disclosure

After the fix is released:

• We publish a security advisory on GitHub
• We update CHANGELOG.md with CVE (if assigned)
• We credit the reporter (unless they prefer anonymity)

1. Keep XylonStudio updated to the latest version
2. Use .env for secrets, never hardcode API keys
3. Limit server access to trusted networks only
4. Enable firewall on production servers
5. Review code before running untrusted designs

1. Never commit secrets (API keys, passwords, tokens)
2. Use parameterized queries for database access
3. Validate all user inputs (see agent/core/input_validator.py)
4. Follow secure coding standards (OWASP Top 10)
5. Run security tests before submitting PR

LLM-generated code is untrusted by default.

XylonStudio runs all generated RTL in sandboxed Docker containers:

• No network access
• Limited filesystem (read-only)
• Resource limits (CPU, memory, time)

See agent/sandbox/ for implementation.

Customer designs are treated as secrets.

• Stored in proprietary/customer-data/ (excluded from git)
• Never used for LLM training without explicit consent
• Encrypted at rest (if enterprise tier)
• Deleted upon request (GDPR/CCPA compliance)

Awareness: Adversarial specs could attempt prompt injection.

Mitigation:

• Input validation (agent/core/input_validator.py)
• Output sanitization (RTL linting before execution)
• Cost limits (agent/core/cost_limiter.py)

If you find a successful prompt injection, please report it!

• ✅ Input validation (length, charset, rate limiting)
• ✅ Sandbox execution (Docker isolation)
• ✅ Cost limiting (prevent runaway LLM costs)
• ✅ Secrets management (.env, never in git)
• ✅ Output sanitization (Verilator linting)

• Formal verification (prove RTL correctness)
• Differential testing (compare against reference)
• Model watermarking (detect stolen models)
• Rate limiting per IP (DDoS prevention)

We recognize security researchers who responsibly disclose vulnerabilities:

(No reports yet - be the first!)

Reward: Public acknowledgment + XylonStudio swag

Security Team: hello@xylonstud.io

PGP Key (for encrypted reports):

-----BEGIN PGP PUBLIC KEY BLOCK-----
[To be added]
-----END PGP PUBLIC KEY BLOCK-----

Response time: 48 hours maximum

CVE Database: GitHub Security Advisories

Past advisories: None yet (project launched 2026-04-01)

Thank you for helping keep XylonStudio secure! 🛡️

Last updated: 2026-04-02