Skip to content

Security: NX1X/OpenMila

SECURITY.md

Security Policy

OpenMila runs on your own machine: it records audio, keeps transcripts and voice profiles on disk, and can paste text into other applications. Security reports are taken seriously.

Reporting a vulnerability

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

Report it privately through GitHub's private vulnerability reporting. Reports are acknowledged within a few days.

Please include:

  • what the issue is and what an attacker could do with it;
  • steps to reproduce, your OpenMila version (openmila --version) and your operating system;
  • a proof of concept, if you have one.

If the problem is in code OpenMila shares with Mila (anything under Mila/ or Packages/), it probably affects Mila on macOS too. Say so in the report; it will be passed on to Mila's maintainers through their own reporting channel once a fix is coordinated.

Supported versions

Fixes go into the newest release. Update before reporting.

Good to know

  • Local transcription never sends audio anywhere. The optional remote backend uploads audio only to the server you configure, and the app says so while it is active.
  • AI features run a command-line tool or an OpenAI-compatible endpoint that you configure. OpenMila ships no API keys.
  • Secrets you enter (API keys, tokens) are stored in files readable only by your user account.
  • The MCP helper is off until you turn it on in Settings, and turning it off takes effect immediately. It is a consent switch, not a sandbox: it runs as your user, like the rest of the app.
  • Updates are checked against this repository's GitHub releases and are never installed without you.

There aren't any published security advisories