Question 1

How do I know if a SKILL.md file I found on GitHub is safe for my classroom agent?

Accepted Answer

You mostly can't today — published research found roughly of agent skills carry at least one vulnerability. SkillzBench is building exactly this check: every skill version is pinned to a commit SHA, license-checked, security-scanned, and quality-judged before receiving a 0-100 trust score.

Question 2

What does 'Lift' mean on a SkillzBench benchmark?

Accepted Answer

Lift is the measured improvement a skill provides, in percentage points. We test every skill in a sandbox with the same model, once with the skill and once without, and publish the difference. It's always shown separately from trust ratings, and money never touches scores.

Question 3

Can SkillzBench help my district meet AI governance requirements?

Accepted Answer

With only 18% of principals reporting district AI guidance, most schools lack policy infrastructure. SkillzBench's Teams plan is being designed for org loadouts, allowlists, and audit trails — giving administrators a control panel over which skills agents can use. Note that SkillzBench is pre-launch; join the waitlist to follow Teams availability.

Question 4

When can I actually start using SkillzBench?

Accepted Answer

SkillzBench is pre-launch and actively being built. The waitlist and newsletter are live now, so joining today puts you first in line for access to Discover, Verify, Benchmark, and Equip as each pillar ships.

Question 5

How is SkillzBench different from just reading GitHub stars and README claims?

Accepted Answer

GitHub stars measure popularity, not safety or effectiveness. SkillzBench independently validates versions, scans for vulnerabilities, checks licenses, and runs controlled benchmarks that publish actual Lift in percentage points — evidence a README can't provide.

Agent Skill Migration for Teachers