Skip to content

SECURITY.md

Latest commit

 

History

History
88 lines (65 loc) · 3.36 KB

SECURITY.md

File metadata and controls

88 lines (65 loc) · 3.36 KB
Raw

Security Policy

Arte Ogre is a local-first, no-telemetry desktop image editor. It still parses untrusted input - image files, native .ogre/.ora documents, SVG, and local WASM/Lua plugins - so we take security reports seriously.

Supported versions

Security fixes land on the latest 1.x release line and master.

Version Supported
1.x ✅
< 1.0 ❌

Reporting a vulnerability

Please do not open a public issue for security problems.

Report privately through GitHub:

  1. Go to the repository's Security tab.
  2. Click Report a vulnerability to open a private advisory.

If you cannot use GitHub Security Advisories, reach the maintainers via visorcraft.com.

Please include:

  • The affected version or commit.
  • A description of the issue and its impact.
  • Steps to reproduce, ideally with a minimal sample file or plugin.

We aim to acknowledge a report within a few days and to keep you updated as we investigate. Please give us a reasonable window to ship a fix before any public disclosure; we're happy to credit you in the release notes.

Scope

In scope:

  • File loaders (ogre-io) - .ogre, .ora, PSD, SVG/SVGZ, EXR, TIFF, PNG, JPEG, WebP, BMP, and other enabled raster decoders. Maliciously crafted files that cause crashes, unbounded memory or CPU use, path traversal, or memory-safety issues.
  • Plugin sandbox (ogre-plugins) - escapes from the wasmtime fuel/memory caps, WASM capability boundary, or Lua sandbox.
  • The engine (ogre-core, ogre-gpu) - memory-safety or correctness bugs reachable from untrusted input.
  • Network boundary - update/model requests that upload document data, accept an unverified model, or write outside their intended cache path.

Out of scope:

  • Availability of the GitHub API or the upstream model host.
  • Third-party dependency advisories already tracked and fixed upstream, unless Arte Ogre exposes additional impact.
  • Issues that require a local attacker who already controls your account and can replace application or plugin files.

Hardening notes

  • ogre-core is built with #![deny(unsafe_code)].
  • The native .ogre loader bounds canvas, buffer, and vector geometry on load, rejecting oversized or malformed manifests instead of allocating blindly.
  • Tile decoding caps the decompressed size of each tile.
  • WASM plugins receive no host imports or WASI capabilities. They run under a fuel limit and a 512 MiB linear-memory cap.
  • Lua exposes only constrained document commands. Filesystem, process, package, dynamic-code-loading, and debug globals are removed; memory is capped at 64 MiB and execution at 10,000,000 instructions.
  • Plugin manifests are capped at 64 KiB and entry values must be contained relative paths with no absolute path or .. component.
  • AI model downloads stream to a .part file, require the expected size and a pinned SHA-256 checksum, then install by atomic rename.
  • Arte Ogre has no telemetry.

Network behavior

Normal editing, plugins, save/export, and model inference are local. Network access happens only after an explicit user action:

  • Help → Check for Updates queries the GitHub API with a short timeout.
  • First use of Refine edges with AI downloads the pinned IS-Net model from GitHub Releases.

Neither path uploads document or image content.