Question 1

How does SkillzBench measure whether an agent skill actually helps?

Accepted Answer

Every skill is tested in a sandbox with the same model, run with and without the skill, and the result is published as Lift in percentage points. Lift is always shown separately from ratings, so you can distinguish measured performance impact from overall trust quality.

Question 2

What does the 0-100 trust score actually include?

Accepted Answer

The trust score combines security, quality, license, maintenance, and community signals into a single 0-100 rating. Money never touches scores, and Lift is always displayed separately, so a well-marketed skill can't buy its way to a high rating.

Question 3

Can SkillzBench scan skills that contain scripts, not just instructions?

Accepted Answer

At launch, SkillzBench covers instruction-only skills — SKILL.md packages plus text references. Script-bearing skills are listed link-out only, so you always know exactly what scope has been verified versus merely referenced.

Question 4

When can my team start using SkillzBench?

Accepted Answer

SkillzBench is pre-launch and actively being built — the waitlist and newsletter are live now, but the product itself is not yet public. Joining the waitlist gets you launch updates and early access as the five pillars roll out.

Question 5

How is SkillzBench different from public leaderboards like SWE-bench or WebArena?

Accepted Answer

Public leaderboards score models and saturate once tuned against them, and results are highly sensitive to model, harness, tools, and evaluation method. SkillzBench instead benchmarks individual agent skills in a controlled sandbox — same model, with and without the skill — publishing Lift in percentage points, plus a trust score covering security, license, and maintenance risk.

Agent Skill Benchmarking for Enterprise Teams