This project is currently pre-1.0 and maintained on a best-effort basis. Security fixes are applied on the latest default branch.
Please report security issues privately and do not create public issues for unpatched vulnerabilities.
Open a private advisory at: https://github.com/Olajide-Badejo/CUDA-Matrix-Library/security/advisories/new
Include:
- A clear description of the issue
- Impact and exploitability assessment
- Steps to reproduce (if possible)
- Suggested mitigation (if known)
- Acknowledge receipt within 3 business days
- Triage and confirm impact
- Prepare a patch and release notes
- Coordinate responsible disclosure timeline
Potential security-sensitive areas include:
- Native extension boundaries (Python bindings)
- Memory allocation and host/device copy paths
- Input shape validation and integer overflow risks