Skip to content

Security: Goldloli/image-studio

Security

SECURITY.md

Security Policy

Image Studio runs image models locally and does not operate a hosted image service. Security reports should still cover the desktop host, IPC boundary, model installation and archive extraction, worker supervision, update/release workflows, and bundled dependencies.

Supported Versions

Security fixes target the latest published 0.1.x release and main. Older pre-release builds may be investigated but are not automatically patched.

Reporting a Vulnerability

Do not open a public issue for an exploitable vulnerability or include API keys, model credentials, private paths, or user images in a report.

Use the repository’s Security → Report a vulnerability private advisory flow:

https://github.com/Goldloli/image-studio/security/advisories/new

Include the affected version or commit, platform, reproduction steps, expected impact, and the smallest sanitized proof of concept. If private reporting is unavailable, open a non-sensitive issue asking the maintainer to establish a private channel without disclosing the exploit.

Scope

In scope:

  • Tauri IPC and capability bypasses;
  • path traversal, symlink/reparse-point escapes, archive extraction flaws;
  • arbitrary command or worker execution;
  • model download integrity or activation bypasses;
  • secret exposure and release artifact tampering;
  • denial of service that survives a normal stop/restart;
  • dependency vulnerabilities with a reachable product impact.

Out of scope without product-specific impact:

  • model-generated content quality or prompt safety;
  • unsigned-build Gatekeeper/SmartScreen warnings already documented;
  • denial of service requiring direct access to a trusted local administrator;
  • vulnerabilities only in optional services that Image Studio does not use.

Please allow time for triage and a coordinated fix before public disclosure.

There aren't any published security advisories