Skip to content

Security: anshulforyou/grandma

Security

SECURITY.md

Security

Model

  • Your memory never leaves your machine. The engine makes no network calls of its own. Model calls happen through your own Claude Code installation and account.
  • No telemetry. Nothing is collected, phoned home, or aggregated.
  • Secrets stay out of memory by policy and by test. The capture doctrine instructs sessions to store pointers (where a credential lives), never values, and the integrity suite greps memory for token patterns on every run.
  • Hooks execute shell commands from your project's .claude/settings.local.json. Grandma only installs hooks pointing at scripts inside this engine, installs are idempotent, and every hooked script carries recursion guards and cost breakers. Review lib/grandma-rehydrate.sh and lib/grandma-session-end.sh; they are short.
  • Your memory home is a git repo you control. Treat it like a diary: keep the remote private if you add one.

Reporting

Open a GitHub security advisory or email the maintainer (see profile). Please do not open public issues for anything involving leaked personal data.

There aren't any published security advisories