Skip to content

Security: cogitave/yuva

Security

SECURITY.md

Security Policy

Yuva is a research-stage operating system project. Honest scope first:

  • There is no SLA and no bug bounty.
  • Nothing here is claimed to be secure or production-ready. The kernel's own boot output states its claim boundary in machine-emitted honesty tokens (e.g. sec=ASSUMED-FROM-LITERATURE, sidechannel=NOT-CLAIMED, key=HOST-CUSTODIED-PER-RUN — custody, not confidentiality).
  • "Verified" means Kani-proven leaf crates plus boot-asserted witnesses, not a whole-kernel functional-correctness proof.

Reporting

Open a GitHub issue for anything you find — including proof gaps, hollow witnesses, or overclaims in prose, which this project treats as seriously as memory bugs. If the finding is sensitive enough that a public issue feels wrong, use GitHub's private vulnerability reporting on this repository (or contact the repository owner directly).

There is no embargo process; fixes land through the normal PR loop with the full CI gate set.

Threat model pointers

  • docs/assumptions.md — the residual trusted computing base the proofs do NOT discharge (hardware, QEMU/KVM, the toolchain, the hand-written assembly), stated assumption by assumption.
  • The honesty-token discipline — every witness line names what is not claimed, and CI rejects boots that overclaim or skip; see the run scripts under scripts/.

There aren't any published security advisories