agent-skill-lock is security-adjacent software. Please report vulnerabilities privately.

Use GitHub's private vulnerability reporting if available, or email the maintainer listed on the GitHub profile.

Please include:

• Affected version or commit.
• Reproduction steps.
• Expected and actual behavior.
• Whether the issue can leak secrets, hide capability drift, or cause unsafe CI behavior.

Do not include real tokens, private keys, or production secrets in a report.

In scope:

• Secret values written into agent-skills.lock.json.
• Absolute local paths or usernames written into lockfile IDs or metadata.
• Nondeterministic output that hides meaningful drift.
• Parser crashes caused by ordinary malformed config.
• CI Action behavior that needs broader permissions than documented.
• CLI output that incorrectly claims a skill, server, or config is safe.

Out of scope:

• A locked malicious skill being malicious. This tool detects drift; it does not certify safety.
• Vulnerabilities in third-party scanners invoked outside the core CLI.
• Private repository configurations not shared in a reproducible report.

Security fixes should lead with user impact, not branding. Patch releases should be small and direct.