Skip to content

Security: Aevorine/Synorive

Security

SECURITY.md

Security Policy

Reporting a vulnerability

Please report privately through GitHub Security Advisories rather than opening a public issue.

Include what you did, what happened, and what you expected. A proof of concept helps but is not required. Expect an initial reply within about a week — this is a small project, not a company with an on-call rota, and saying so plainly is better than promising an SLA that will be missed.

Supported versions

Only the latest release receives fixes. There are no long-term support branches.

What this project already does about security

Stated so you can check the claims rather than take them on faith:

  • The engine binds to 127.0.0.1 only. It is not reachable from the local network unless you deliberately change that.
  • The browser-render service also binds to loopback only, and rejects any URL that is not http/https — otherwise anything able to reach that port could read local files through file:///.
  • The PDF print window runs with nodeIntegration: false, javascript: false, sandbox: true. The HTML it renders contains text fetched from the public web.
  • Web search and cloud inference are separate switches. Turning off one does not silently leave the other on.
  • The clipboard image overlay never goes to the network, regardless of settings.
  • Cookies injected for authenticated capture go into a per-lane in-memory partition, are cleared before each capture, and never touch the default session or disk.
  • Android update downloads hard-reject any non-GitHub host and verify the downloaded byte count against the size GitHub reports.

Known gaps — stated rather than glossed over

  • The Windows builds are not code-signed. No certificate has been purchased, so Authenticode verification is skipped and SmartScreen will warn about an unknown publisher. Desktop update integrity therefore rests on the sha512 in latest.yml served over HTTPS, not on a signature. Do not treat the desktop update channel as authenticated.
  • API keys you enter (Brave, Serper, Tavily, Exa, Semantic Scholar, cloud inference) are stored locally in your user data directory. They are not encrypted at rest beyond the file permissions your OS gives that directory.
  • Self-hosted SearXNG is a third party in the loop. If you point it at a public instance instead of your own, that instance sees your queries.

There aren't any published security advisories