Only the latest release on the main branch receives active security updates and patch releases.
| Version | Supported |
|---|---|
| main | ✅ |
| < 0.1 | ❌ |
Do not report security vulnerabilities through public GitHub issues, discussions, or pull requests.
If you discover a potential security vulnerability, memory leak, tensor-level constraint exploit, or execution failure within immortal-inference, please submit a report privately via email: theimmortalpythonlabs@proton.me
- Response Expectation: Initial acknowledgement within 48 hours, followed by patch status updates.
To help us investigate and resolve the issue quickly, please include:
- A clear description of the vulnerability or boundary bypass.
- Minimal reproduction code or environment details (PyTorch version, model architecture, logit processor configuration).
- Any potential impact or proof-of-concept payload.
immortal-inference enforces mathematical containment and logit-level restrictions during inference execution.
- Out of Scope: Stochastic model hallucinatory outputs where logit masks are explicitly bypassed or unconfigured by the user application.
- In Scope: Token-masking leaks, runtime execution escapes, illegal state transitions past active Pāṇinian constraint boundaries, and memory/buffer corruption within the core C++/Python binding layer.
Thank you for helping keep non-stochastic inference safe and reliable for everyone.