Skip to content

Security: bext-stack/bext

Security

SECURITY.md

Security Policy

Supported versions

bext is pre-1.0. Security fixes land on the latest published version (0.2.x at time of writing). Older versions do not receive backports. If you're on an older release, upgrade first and re-test before filing a report.

Version Supported
0.2.x
0.1.x

Reporting a vulnerability

Please do not file public issues for security vulnerabilities. Public disclosure before a fix is available puts every bext user at risk.

Preferred: GitHub Security Advisories

  1. Go to https://github.com/bext-stack/bext/security/advisories/new
  2. Fill out the template — be specific about affected crates, versions, reproducer, and impact.
  3. GitHub will notify the maintainer and open a private discussion thread.

This channel is encrypted end-to-end between you and the maintainer, and GitHub handles coordination with the CVE database once a fix ships.

Alternative: email

If you can't use GitHub advisories, email security@bext-stack.dev with:

  • Affected crate(s) and version(s)
  • A minimal reproducer (ideally a cargo test-style POC)
  • Your assessment of impact (auth bypass? RCE? DoS? information disclosure?)
  • Whether you want public credit once disclosed
  • Your preferred disclosure timeline

PGP is not required but welcomed. The key fingerprint will be published here once a maintainer key is generated.

What to expect

Phase Timeline What happens
Acknowledgement within 48 hours You get a reply confirming receipt and a tracking ID
Triage within 5 days Severity assessment, affected-version analysis, CVSS score
Fix development 1-30 days Depends on complexity and severity
Coordinated disclosure 90 days max Public advisory + CVE + patched release

For critical severity issues (auth bypass, RCE, data loss) we aim for a patch within 7 days. For medium severity, 30 days. For low severity, bundled into the next regular release.

Scope

In scope:

  • All published bext crates on crates.io and npm (the 13 phase-1 crates, and any crates added in phase 1.5 / phase 3).
  • Published container images on GHCR (ghcr.io/bext-stack/bext:*).
  • Subtree mirror repos (bext-stack/bext-plugin-api, bext-stack/bext-nginx-compat).

Out of scope:

  • Third-party plugins that use bext-plugin-api. Report to the plugin author, not to us.
  • Vendored upstream dependencies (turbopack, react-compiler, tsc_rs). Report to those projects directly.
  • Deployment misconfiguration — if someone leaves their license key in a public bucket, that's not a bext vulnerability.
  • Social engineering of maintainers or users.
  • Denial of service via resource exhaustion unless it bypasses the existing rate-limit / WAF configuration in a way that can't be stopped with config.

Disclosure policy

  • We follow coordinated disclosure: we agree on a public disclosure date with the reporter, ship the fix, and publish the advisory simultaneously.
  • If you publicly disclose before a fix is available, we will still ship the fix, but we can't give advance notice to other users.
  • If we can't reach you after the 90-day window, we reserve the right to disclose unilaterally.

Credit

We credit security reporters in the advisory and release notes by default. If you want to remain anonymous, say so in the initial report.

Hall of fame

(empty so far — be the first!)


For plugin authors

If you publish a bext plugin:

  1. Treat bext-plugin-api as a public ABI and watch for its security advisories.
  2. Your plugin's host crate (bext-plugin-wasm, bext-plugin-quickjs, bext-plugin-nsjail) provides isolation, but a bug in your plugin can still leak data through host function calls — audit accordingly.
  3. Report vulnerabilities in the plugin API surface via the same channels above. Vulnerabilities in your own plugin code are your responsibility.

For commercial / enterprise users

If you run bext in production and need a guaranteed response time faster than 48 hours, reach out via the commercial licensing channel (open an issue tagged licensing) to discuss a support contract.

There aren't any published security advisories