Skip to content

Security: 2lab-ai/zauthenticator

SECURITY.md

Security policy

zauthenticator holds TOTP secrets, so it is built to keep a few guarantees: extension pages and its service worker open no network connections and load no remote resources, its content script makes no requests either, the extension opens no pages other than its own, it runs no remote or dynamic code, secrets are encrypted at rest, and auto-fill gives a code, and only a code, to the exact sites you allowed. A break in any of them is a vulnerability.

Reporting

Report privately through GitHub's private vulnerability reporting. On the repository's Security tab, choose Report a vulnerability, or open https://github.com/2lab-ai/zauthenticator/security/advisories/new.

Do not open a public issue or pull request for a vulnerability. Include the version or commit, the browser and its version, the steps to reproduce, and what you observed.

Reports are handled on a best-effort basis. A fix ships as a new release, and the advisory is published once that release is available.

Scope

  • The extension in extension/ and the release zips built from it.
  • The trust gate (scripts/check-policy.mjs), the end-to-end test, and the CI and release workflows, because they are what lets anyone check the guarantees.

Only the latest release is supported.

What counts

  • Any network egress from the extension: a request of any kind to any host, including through a CSP gap, a browser API that bypasses connect-src, a request made from the content script inside a web page (where the extension's CSP does not apply), or a navigation the extension triggers.
  • Opening a page outside the extension: a tab, window or navigation to anything other than the extension's own pages, for example through a navigation sink that npm run check misses.
  • Secret exposure: a TOTP secret, the passphrase or the vault key that is written unencrypted to chrome.storage.local or chrome.storage.sync, included in an export, made readable outside the extension's trusted contexts (its pages and service worker; the content script and web pages are outside), or recoverable from the stored record without the passphrase.
  • Auto-fill to the wrong place: a code delivered to a page or frame whose exact origin you did not bind (a look-alike or a subdomain included), a code delivered while the vault is locked or idle, anything other than the code (a secret, an account name, the key) reaching a page, a binding stored or site access granted without your Allow and the browser's own prompt, the content script running on a site that is not bound, or a form submitted by the extension.
  • CSP bypass or code injection: running remote, inline or dynamically evaluated code in an extension page, for example through a crafted otpauth:// URI, a QR code or a backup file.
  • Cryptographic weaknesses: IV reuse, weakened key-derivation parameters, missing authentication of the ciphertext.
  • Gate or supply-chain bypass: a change that breaks a guarantee but still passes npm run check, a vendored file that does not match vendor/SHA256SUMS, or a release zip whose contents differ from extension/ at the tagged commit.

Out of scope

  • An attacker who already controls your operating system account, browser or browser profile, or who uses your session while the vault is unlocked.
  • Weak or reused passphrases.
  • Wrong codes caused by a wrong system clock. The system clock is the only time source.
  • Codes read from the system clipboard after you copy them.
  • A bound site reading the code that was filled into it, which is what auto-fill is for, including scripts injected into that site.
  • Field-detection misses: a bound page whose code field is not found, or where a different empty field of that page is filled. Detection is a heuristic; report these as ordinary bugs.

There aren't any published security advisories