Skip to content

Security: omm-hippo/omm

SECURITY.md

Security Policy

Supported Versions

omm follows a rolling-release model and does not maintain long-term-support branches.

Version Security support
Latest published release Supported
Older published releases Not supported
Current main branch Reports welcome; this is unreleased development code

Official distribution channels can temporarily provide different versions. Include omm --version and the installation method in a report, and compare them with the latest GitHub release.

Reporting a Vulnerability

Do not open a public GitHub issue, discussion, or pull request for a suspected vulnerability. The repository does not currently enable GitHub's private vulnerability-reporting form.

Email omm.hippo@gmail.com with the subject [Security] and include:

  • A concise description and the potential impact.
  • The affected omm version or commit and installation method.
  • Operating system, architecture, Python or Node.js version, and affected local runner where relevant.
  • Minimal reproduction steps or a proof of concept.
  • Redacted logs or screenshots that help confirm the behavior.

Do not send access tokens, service-account credentials, unredacted private paths, or third-party personal data. Do not test against systems, accounts, services, or data you do not own or have permission to assess. Avoid high-volume testing against public or quota-limited services.

The maintainers aim to acknowledge a report within a few days, confirm the affected scope, prepare and verify a fix, and coordinate disclosure timing. If you are unsure whether a report is security-sensitive, use the private email path.

Scope

Reports are especially useful for vulnerabilities involving:

  • Model download, installation, update, import, linking, cleanup, and file integrity.
  • Official installation scripts and Python, npm, Homebrew, or portable release supply chains.
  • Trusted commits, signed recommendation catalogs, checksums, and signature validation.
  • Archive extraction, path handling, permissions, and cross-volume copies.
  • Local configuration, credentials, tokens, or sensitive filesystem state.
  • Benchmark, telemetry, error-report, Firebase, localfit-server, or cf-worker authorization and privacy boundaries.
  • Supported local AI runner integrations where omm creates or amplifies the security impact.

Issues that exist entirely in an upstream model or runner should normally be reported to that upstream project. If omm's integration changes the impact or trust boundary, include that detail in a private report to this project too.

Disclosure and Fixes

Please allow reasonable time for investigation and a coordinated fix before publishing technical details. Security fixes are supported only when the patched release or destination artifact has actually been published and verified; a pull request or successful CI run alone is not a released fix.

Non-security Bugs

For ordinary bugs, installation problems, and compatibility requests without a security impact, use the public GitHub issue templates.

There aren't any published security advisories