skill-vetter

To vet AI skills for safe installation, run a standardized protocol that evaluates source credibility, reviews code for unsafe patterns, and assesses permission scope to generate a machine-readable risk report.

A skill vetting protocol is a security-first assessment process for AI skills. You need it before adding any new skill from ClawdHub, GitHub, or shared agent-skills to prevent unsafe or unvetted installations.

You check AI skill permissions by evaluating read/write needs, network access, and command requirements. This assessment assigns a risk level and identifies red flags to determine an installation verdict.

Yes, the vetting process works with skills from GitHub, ClawdHub, and shared agent-skills. It assesses origin, author reputation, and update history across these sources to gauge trust before installation.

The mandatory code review requires reading all files to identify red flags and unsafe patterns. Skills are rejected based on explicit criteria found during this mandatory review and permission scope evaluation.

A machine-readable vetting report includes identified red flags, assigned risk levels, and a final installation verdict. It details found issues and recommended actions to verify safe skill deployment.