Please don't open a public issue for a security problem. Report it privately through GitHub instead: on this repository, open Security → Report a vulnerability. Only the maintainers can see the report.
Include what you found, how to reproduce it, and what an attacker could do with it. Use fictional data in your report, as in CONTRIBUTING.md: never a real person's contact details or anything from a real Sellbird database.
We'll acknowledge the report within a few days, keep you updated while we work on a fix, and credit you in the release notes unless you'd rather we didn't.
Security fixes go into the latest release on main. Sellbird is updated with
git pull, so please check the problem still happens there.
Sellbird runs on your own machine with your own credentials. The most important boundaries are:
- the dashboard's sign-in token and its loopback-only listener (docs/dashboard.md);
- the isolation of model calls, which run with no tools (docs/how-it-works.md);
- the Outbox holds that stop planted instructions in websites or replies from sending email nobody approved (docs/approval-and-sending.md).
A problem that needs code already running as your own user account is out of
scope: that code can read .env and the database directly.
The hosted service at sellbird.ai is a separate product; report problems with it the same way and say it's about the hosted service.