Skip to content

Security: mralexivy/AutoMode

Security

SECURITY.md

Security policy

AutoMode is a Claude Code mod. A mod is code that runs with the permissions of the person who installs it, and it is not sandboxed. We take reports about it seriously.

What AutoMode does

You can list what AutoMode does without running it:

claude plugin validate ./AutoMode

The README explains each call in Review it before you install it. In short:

  • It makes no network calls of its own. It has no $.http.fetch and no MCP calls.
  • It writes only to its own data folder, ~/.claude/plugins/data/automode/, and, with debugLog on, to a log in the plugin folder.
  • It starts four commands: chmod and rm -f on its own data files, and the read-only git rev-parse and git status.
  • It never stores or sends the value of ANTHROPIC_API_KEY. It only tests whether the variable is set.

Report a vulnerability

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

  1. Go to the Security tab of this repository and choose Report a vulnerability. This opens a private advisory that only the maintainers can read.
  2. Say what you found, how to reproduce it, and which version you tested. Add the output of /automode debug.

We aim to reply within seven days. We will tell you when a fix is released, and we will credit you in the release notes if you want that.

Supported versions

Only the latest release gets security fixes.

What counts

Examples we want to hear about:

  • A path where AutoMode writes outside its own folders.
  • A way for a prompt, a tool result or a file to make the guardian store a secret in plain text.
  • A way to make the git guard or the subagent guard allow a destructive command.
  • A secret, a token or private data in this repository.

Not in scope

  • Claude Code's own behavior. Report that to Anthropic.
  • A mod that you installed from another author.
  • The fact that a mod runs with your permissions. That is how mods work.

There aren't any published security advisories