Question 1

How does SkillzBench audit an agent skill for security risk?

Accepted Answer

Each skill is pinned to a specific commit SHA, validated, license-checked, security-scanned, and quality-judged. The results roll into a trust score from 0-100 built on security, quality, license, maintenance, and community signals — and the sandbox benchmark Lift is always shown separately from the rating.

Question 2

Can SkillzBench scan skills with embedded scripts?

Accepted Answer

At launch, SkillzBench's deep verification and benchmarking covers instruction-only skills — SKILL.md packages plus text references. Script-bearing skills are listed link-out only, so you always know exactly what category a skill falls into before you evaluate it.

Question 3

Why should a security team trust SkillzBench's scores over a skill author's claims?

Accepted Answer

SkillzBench is an independent testing lab: it benchmarks every skill in a sandbox with the same model, with and without the skill, and publishes the Lift in percentage points. Money never touches scores, and Lift is never blended into ratings — so the evidence stands on its own.

Question 4

When can I get a skill audit and risk report for my agents?

Accepted Answer

SkillzBench is pre-launch and actively being built. The waitlist and newsletter are live now — joining puts you first in line as the Discover, Verify, Benchmark, and Equip pillars roll out, including per-agent loadouts and the MCP server.

Question 5

How is this different from just reading a skill's README on GitHub?

Accepted Answer

A README tells you what an author claims; SkillzBench tells you what independent testing found. You get version pinning to a commit SHA, license and security scans, sandbox benchmark Lift data, continuous upstream change tracking with re-scan notifications, and least-privilege per-agent loadouts — none of which GitHub browsing provides.

Skill Audit & Risk Report for Designers