SigSieve is security software and has not been independently audited. Its reviews so far are internal AI-assisted reviews in separate contexts, alongside automated tests. See the evidence and remaining gaps and findings register. We take reports seriously and appreciate responsible disclosure.

Fixes target main and the next release. Backports to older releases are not promised, and no long-term support window is promised. Include the exact commit when reporting an unreleased build.

Please do not open a public issue for security problems.

Report privately through GitHub's private vulnerability reporting ("Security" tab → "Report a vulnerability").

If that form is unavailable, open a public issue titled "Security contact request" without any details of the problem; the maintainer will reply with a private channel.

Please include:

• affected version or commit,
• a description of the issue and its impact,
• steps or a proof of concept to reproduce (synthetic keys only, never real private keys or credentials),
• any suggested fix.

• Acknowledgement within 7 days.
• An initial assessment within 14 days.
• Coordinated disclosure: we aim to fix confirmed issues and publish an advisory within 90 days, crediting you unless you prefer otherwise.

In scope: anything that lets a downstream client list or use a non-allowlisted identity, reach the upstream agent with operations other than allowlisted listing and signing, crash or exhaust SigSieve, make SigSieve delete or replace files it didn't create, or make sigsieve connect forward an unfiltered agent. Also in scope: documentation that claims a guarantee the code doesn't provide.

Out of scope: issues that require a compromised local user account, root, or a compromised upstream agent; misuse of explicitly allowed keys; and the limitations documented in docs/THREAT_MODEL.md and docs/LIMITATIONS.md. We still welcome ideas for hardening in these areas as regular issues.