Skip to content

Security: stonefold-ai/spec

SECURITY.md

Security policy

Stonefold is an enforcement layer and this repo is the canonical home of its specifications. A flaw in the spec that would let a conformant implementation be bypassed or exploited is a security issue, not an ordinary defect report — please report it privately.

How to report

Do not open a public issue or PR. Use GitHub's private vulnerability reporting on this repository (Security tab → "Report a vulnerability"). If that is unavailable to you, email gallas.robert@gmail.com with the subject line STONEFOLD SECURITY.

A useful report includes:

  • The RFC section (§n), schema, or fixture affected.
  • A reproduction. The ideal shape is an intent + policy + registry that demonstrates the hole — an implementation could follow the spec exactly and still allow the action.
  • The stated guarantee you believe is broken.

Flaws in the reference gateway's code (as opposed to the spec text) are handled under the same policy in the stonefold repo; either entry point reaches the same maintainer.

Reports must be verified by a human before sending. Unverified AI-generated reports waste the capacity this policy exists to protect and will get the sender banned.

What to expect

This is a solo-maintained project, so the promises are modest but honest:

  • Acknowledgment within a few days.
  • An honest assessment of whether the report is valid, and why.
  • Coordinated disclosure: a fix (normally a CS-nnn change-set item) or a documented limitation before the report is made public, on a timeline agreed with you.
  • Credit in the changelog and advisory, if you want it.
  • There is no bounty program.

Scope

In scope: spec wording, schemas, or fixtures that would make a conformant implementation exploitable — enforcement bypasses, scope or identity injection, kill or audit guarantees that don't hold as written, under-specified corners with a demonstrated attack.

Out of scope: implementation bugs where the spec is correct (those go to the stonefold repo), and purely editorial issues (those are ordinary public issues, and welcome).

There aren't any published security advisories