Question 1

What does 'Lift' mean in SkillzBench's benchmarking?

Accepted Answer

Lift is the measured performance difference, in percentage points, between running the same model in a sandbox with and without a given skill. Because both runs use the same model and scaffold, the comparison is controlled — unlike leaderboard results that mix scaffold choices and can show a 33x cost spread on identical tasks.

Question 2

Can I trust the trust score, or is it pay-to-play?

Accepted Answer

The trust score is a 0-100 rating built from security, quality, license, maintenance and community signals — and money never touches scores. Lift is always shown separately from ratings, so commercial relationships can't inflate either number.

Question 3

Does SkillzBench support script-bearing skills?

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 can find them but they aren't verified or benchmarked in the sandbox until the lab expands scope.

Question 4

When can I actually use 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 early access as the Discover, Verify, Benchmark and Equip pillars come online.

Question 5

How is this different from just browsing GitHub for skills myself?

Accepted Answer

GitHub gives you raw files with no validation: no commit-SHA pinning, no security scan, no license check, no benchmark. SkillzBench continuously crawls and dedupes skills, verifies every version, runs with/without sandbox benchmarks, and lets you equip agents through per-agent loadouts — plus it re-scans and re-benchmarks when upstream skills change.

Agent Skill Benchmarking for Developers