Question 1

What exactly does SkillzBench security-scan on an agent skill?

Accepted Answer

Every SKILL.md instruction package is pinned to a commit SHA, validated, license-checked, security-scanned, and quality-judged before it receives a trust score from 0-100. That score draws on security, quality, license, maintenance, and community signals, so you can see risk at a glance before a skill ever reaches an agent.

Question 2

Can a risky skill affect my sales agents without me knowing?

Accepted Answer

That's the core problem SkillzBench is built to solve — published research found roughly of agent skills carry at least one vulnerability. With SkillzBench's per-agent loadouts, an agent only sees the skills and tools you explicitly toggled on, and every one of those has been scanned and verified first.

Question 3

How do I know a skill actually helps instead of just existing?

Accepted Answer

SkillzBench benchmarks every skill in a sandbox with the same model, run with and without the skill, and publishes the Lift in percentage points. Lift is always shown separately from ratings, and money never touches the scores — so performance claims are measured, not marketed.

Question 4

When can my team start using SkillzBench?

Accepted Answer

SkillzBench is pre-launch and actively being built right now. The waitlist and newsletter are live today, and joining gets you early access as the Discover, Verify, Benchmark, and Equip pillars roll out. At launch, instruction-only skills (SKILL.md plus text references) are fully supported, with script-bearing skills listed link-out.

Question 5

How is this different from just vetting skills manually on GitHub?

Accepted Answer

Manual vetting doesn't scale and doesn't stay current. SkillzBench continuously crawls and dedupes skills from GitHub, pins every version to a commit SHA, runs security and license checks automatically, re-scans and re-benchmarks when upstream changes, and notifies you — plus it delivers everything to your agents through one MCP server with per-agent loadouts.

Agent Security Scanning for Sales Teams