Question 1

How does SkillzBench measure whether an agent skill actually improves performance?

Accepted Answer

SkillzBench is building sandbox benchmarking that runs every skill with the same model, both with and without the skill, and publishes the Lift in percentage points. This differential approach isolates the skill's actual contribution, which research shows is critical since efficiency regressions often can't be explained by prompt length alone. Lift is always shown separately from ratings so scores stay honest.

Question 2

Does SkillzBench support script-bearing skills or only instruction-only SKILL.md packages?

Accepted Answer

At launch, SkillzBench's scope is instruction-only skills — SKILL.md files plus text references. Script-bearing skills will be listed link-out only, so you can still discover them but they won't be benchmarked or equipped through the platform initially. This focused scope keeps validation and security scanning rigorous.

Question 3

How does SkillzBench handle security risk in third-party agent skills?

Accepted Answer

Every skill version is pinned to a commit SHA and goes through security scanning, license checking, validation, and quality judging. Research has found roughly of skills carry at least one vulnerability, and these signals feed into a 0-100 trust score alongside quality, license, maintenance, and community signals.

Question 4

Can I control which skills each of my agents uses?

Accepted Answer

Yes — the Equip pillar is one MCP server plus per-agent loadouts. You toggle skills and tools per agent, and the agent only sees what was chosen. This directly addresses the research finding that retrieval precision collapses as skill libraries grow, by keeping each agent's visible skill pool tight and intentional.

Question 5

How is SkillzBench different from just browsing skills on GitHub or a community marketplace?

Accepted Answer

GitHub gives you raw, unverified SKILL.md packages with no way to compare effectiveness. SkillzBench is an independent testing lab: it continuously crawls and dedupes skills, pins versions to commit SHAs, security-scans and license-checks them, benchmarks them with and without the skill, and keeps you current with re-scans and notifications when upstream changes. Money never touches scores, so verification stays independent of any marketplace incentive.

Verified Agent Skills for Job Seekers