Skip to content

Security: mollendorff-ai/ref

Security

SECURITY.md

Security Policy

Supported versions

Only the latest release gets security fixes. ref update installs it.

Reporting a vulnerability

Please do not open a public issue for a security problem.

Email [email protected] with:

  • what you found and what it lets someone do,
  • steps that reproduce it (the command, MCP call or page involved),
  • the Ref version (ref --version), your operating system and your Chrome version.

You can also use GitHub's private vulnerability reporting, under the Security tab of this repository, if it is enabled.

Ref is a small project kept up on a best-effort basis. I aim to reply within 14 days, and I will tell you what I plan to do about the report. Please give me reasonable time to ship a fix before you share the details publicly. I am glad to credit you in the release notes if you want that.

What is in scope

  • ref fetch, ref pdf and the other commands: how they handle URLs (only http and https are accepted), redirects and the content they read.
  • The MCP server (ref mcp), including what its tools can write to disk.
  • The saved Chrome profile used by ref login (~/.ref/chrome-profile), which holds your signed-in sessions: how Ref creates, uses and protects it, and anything that could leak it.
  • ref update, which downloads a release from GitHub and checks its SHA-256 before installing.

What is not in scope

  • Chrome itself, including the calls it makes to Google: report those to Chromium.
  • A web page that tries to give instructions to the AI agent reading Ref's output. That is a risk of letting an agent read the web. Ref cannot judge what a page means, so treat what it returns as untrusted data.
  • Wrong or incomplete page content is a bug, not a security problem: please open a normal issue for those.

There aren't any published security advisories