CortexWard is a security tool, and we hold it to the standard it enforces on others. This document covers how to report vulnerabilities and the security model of the project itself.

Please do not open public issues for security vulnerabilities.

Report privately via GitHub's private vulnerability reporting ("Report a vulnerability" on the Security tab), or email the maintainers at the address in the repository metadata.

Please include:

• affected version / commit,
• a description and impact assessment,
• reproduction steps or a proof of concept,
• any suggested remediation.

We follow coordinated disclosure aligned with ISO/IEC 29147. We aim to acknowledge reports within 3 business days and to provide a remediation timeline within 10 business days. We will credit reporters who wish to be named once a fix is released.

CortexWard is pre-1.0. Security fixes are applied to main. Once 1.0 ships, this table will track supported release lines.

CortexWard analyzes code that must be treated as untrusted, adversarial input. Our design defends against, at minimum:

• Prompt injection embedded in source, comments, or documentation. Analyzed content is passed to models strictly as data; no tool exists that lets a model approve its own finding.
• Code execution during analysis. Static analysis never runs project build steps. Dynamic verification runs only inside a sandbox with deny-by-default egress.
• Sandbox escape. Progressive isolation tiers (container → microVM) and ephemeral environments between runs.
• Secret handling. Local-only operation is supported; when models are used, egress is explicit and secrets are redacted before any external call.
• Supply chain. The tool's own dependencies are minimal, pinned, and scanned in CI (pip-audit, gitleaks).

See ARCHITECTURE.md §5.1 for the full threat/defense matrix. A complete STRIDE model is developed in Phase 5.

CortexWard is for authorized security testing only. Do not run it against code you do not own or lack permission to test. If it discovers previously unknown vulnerabilities in third-party software, follow coordinated disclosure before publishing details.