Please report a security problem privately, in one of two ways:

• by e-mail to security@larmor.org;
• through GitHub's private vulnerability reporting: on the repository's Security tab, choose Report a vulnerability. Only the maintainers see the report, and you can follow and discuss it there until a fix is released.

Please do not open a public issue for it.

Say what the problem is, how to reproduce it, and which Larmor version and operating system you use (lar --version prints it). Once a fix is released, the report is published as a security advisory, crediting you if you wish.

Fixes go into the next release. Until 1.0, only the latest release is supported.

Most of Larmor reads and writes files on the user's own computer. The part that accepts requests from other programs is the viewer's MCP server, which lets an agent drive BrainViewer:

• it is off until the user starts it (lar bv --mcp, or Serve agents in Settings);
• it listens at 127.0.0.1 only;
• every request needs a 32-byte random token, kept in a file readable by the user only;
• requests whose Host is not the viewer's own address, or that come from a web page (Origin), are refused;
• an agent with the token can read any file the user can read, by loading it, and write screenshots, videos and sessions wherever the user can write; it cannot run other programs.

The security page of the documentation describes this in full. A way around any of these limits is a vulnerability; so is a file that makes a reader run code or write outside the paths it was given.