pkgward is a security tool — vulnerabilities in the engine itself, in the detonation sandbox, or in the intel-pack loader could let attackers evade detection or pivot through deployments. We take these seriously.

Please do not file public GitHub issues for active vulnerabilities.

Instead, email security@cyfar.ca with:

• A clear description of the issue
• A proof-of-concept or repro case (a synthetic malicious package, a crafted intel pack, a fuzzed input — whatever demonstrates the problem)
• Affected pkgward version(s)
• Your contact info (so we can ask follow-up questions; anonymity is OK if you'd prefer)

We'll acknowledge receipt within 72 hours, give you a tracking handle, and keep you posted on remediation timing. Once a fix is released, we'll coordinate a public disclosure window with you.

• Engine bugs that produce a false-negative on a malicious package, where the failure is exploitable (e.g., a malformed archive that crashes the scanner mid-extraction, leaving findings unrecorded)
• Detonation sandbox escapes (a package that breaks out of the Docker container)
• Intel-pack loader bugs that let a crafted overlay corrupt or override baseline detection unsafely
• Anything that lets an attacker write arbitrary content to disk via the scan-queue / archive-extraction path
• Credential exposure in logs, error messages, or persisted DB rows

• False positives on legitimate packages. Open a regular issue; we'll triage.
• False negatives on packages where no current rule covers the technique. Detection coverage is best-effort and depends on the loaded intel pack. Submit a YARA rule PR.
• DoS against the scanner via malformed input that the scanner correctly rejects. Crash-safe error handling is good; making the scanner slower against trivially-detected garbage is generally a fact of life.
• Vulnerabilities in dependencies you can already see via pip-audit / osv-scanner. We treat these as regular bugs and roll fixes in normal releases.
• Issues with the maintainer's private intel pack content. That content is not in this repo; reports about it should go to the maintainer directly.

This file tracks resolved security reports. Empty so far.

If you need to send encrypted reports, request a current PGP key in your initial (unencrypted, content-free) email and we'll respond with a fresh fingerprint.