Skip to content

Security: Adeolu05/drainproof

SECURITY.md

Security policy

Drainproof is security tooling, so we treat defects in it as security issues. This includes a scenario that passes when it should fail, a decision path that fails open, or a redaction gap.

Reporting a vulnerability

Please do not open a public issue for a vulnerability. Use GitHub's private vulnerability reporting ("Report a vulnerability" on the repository's Security tab), or contact the maintainers privately. Include:

  • the affected package and version, or the commit;
  • a minimal policy, scenario or proposal that reproduces the problem;
  • the expected and the observed behaviour.

We aim to acknowledge reports within 5 working days.

What counts

  • Any input that makes the engine return ALLOW when a documented rule requires BLOCK or REQUIRE_APPROVAL.
  • Any error path that produces ALLOW (a fail-open path).
  • An approval grant that is accepted for a proposal or policy version it was not issued for.
  • Concurrency, idempotency or lifecycle bugs that let reserved spending exceed a limit.
  • The runner reporting PASS for a scenario that was not actually evaluated.
  • Secrets appearing in reports, logs or the dashboard.
  • The NEAR executor signing anything outside the documented conditions.

Every confirmed security bug is fixed together with a regression test, and the fix is not considered complete until that test exists.

Handling secrets

  • Never put private keys, seed phrases or API tokens in policies, scenarios or the repository.
  • The NEAR executor accepts a testnet key only through environment variables. Never use a mainnet key with it.
  • CI workflows use contents: read permissions, and third-party actions pinned to commit SHAs.

Scope limits

See docs/threat-model.md and docs/limitations.md for what Drainproof does and does not protect against.

There aren't any published security advisories