Security policy #
Only the latest release receives security fixes. If you run an older release, upgrade before you report a problem that the latest release already fixes.
Report a vulnerability #
Send the report to admin@doubledecimal.com. Do not open a public issue, and do not post the details in a chat room or on social media.
Include:
- The version and the commit from the admin page or
/api/health. - The deployment: Docker Compose, bare Node, or Cloudron.
- The browser and the operating system.
- The steps to reproduce the problem.
- The impact: what an attacker can read, change, or stop.
- A proof of concept, if you have one.
Do not include a real manuscript, a password, a session cookie, a secret link, or the signing secret. Use your own instance and made-up words.
Rules for testing #
- Test only an instance you own or have written permission to test.
- Do not access, change, or delete another person's data.
- Do not run a denial-of-service test against a shared instance.
- Stop when you have enough evidence to show the problem. Do not keep going to see how far it reaches.
- Tell me before you publish anything, so I have time to ship a fix.
What to expect #
I will acknowledge the report, investigate it, and tell you whether it is a bug and what I plan to do. This is a small project, so I cannot promise a fixed timeline. Tell me if you want credit in the release notes.
Scope #
In scope:
- The application code in this repository.
- The Docker and Cloudron packaging.
- The backup, restore and account commands in
scripts/ops.mjs. - The documentation, when it gives an operator unsafe advice.
Out of scope:
- A vulnerability in a dependency that the app does not expose. Report it upstream, and tell me if the app needs a workaround.
- A problem that needs an already-compromised server, an already-stolen password, or an already-stolen session cookie.
- A missing hardening header with no demonstrated impact.
- A secret link that was posted in public by its owner.
- A self-hosted instance that runs an old release with a known fix.
The security model describes what the app stores, what the operator can read and where the trust boundaries are.