Surfguard is a security control — an SSRF address guard — so classification bugs are security vulnerabilities, not ordinary defects. Please report them privately.

Report privately via GitHub's vulnerability reporting: Report a vulnerability (the repo's Security tab → Report a vulnerability).

Open-source reports aren't bounty-eligible, but we also accept them via hackerone.com/basecamp if you'd like the report on your HackerOne record — see the 37signals security response policy.

Do not open a public issue for security bugs.

Anything that lets a blocked address through, for example:

• An address in a blocked range that Surfguard classifies as public.
• A resolution path (encoding, embedding, transition mechanism) that reaches a blocked address despite validation.
• A parser or resolver discrepancy that lets a caller following a documented API contract connect to a different address than Surfguard classified.

Surfguard does not perform the fetch itself. Scheme restrictions, redirect handling, connection pinning, custom DNS64 prefixes, and nonstandard NSS sources have caller or deployment requirements documented in the README. Those boundaries by themselves are not classification bugs, but a bypass when the documented contract is followed qualifies.

Synchronous name resolution has no independent cancellation deadline; callers own request and worker deadlines. Public-addressed internal routes, custom DNS64, ISATAP/6rd, provider aliases, proxy configuration, and resolver/NSS divergence remain deployment risks requiring egress controls.

SG-02 is accepted as a High residual risk: one principal can initiate and approve a release, self-review remains enabled, and the current admin/write radius is external trust. Until a second authorized principal is available, the compensating controls are immutable release tags, no admin environment bypass, protected workflows, least-privilege jobs, reproducible package comparison, canonical-registry-byte confirmation, and read-only control-state checks. This decision is revisited only when a second authorized principal becomes available.

Recovery dispatched from main retains main-bound provenance, and release tags remain unsigned; both are documented accepted residuals. GitHub-hosted runners, the Actions artifact service, and GitHub's control plane remain external trust; fresh-runner verification, reproducible rebuilding, and canonical-registry digest equality reduce but cannot eliminate that exposure. Explicitly overriding a supported Ruby with a vulnerable resolv gem is unsupported residual risk.

Non-security bugs and questions belong in the regular issue tracker.

The latest released version. Fixes ship as a new release, not backports.

We'll acknowledge your report, work with you on a fix, and coordinate disclosure through a GitHub Security Advisory. Reporters are credited in the advisory unless they prefer otherwise.