Question 1

How does SkillzBench measure whether an agent skill actually improves my AI coding agent's output?

Accepted Answer

SkillzBench benchmarks every skill in a sandbox using the same model, running the task with and without the skill, and publishes the difference as a Lift in percentage points. This means you see measured performance impact rather than star counts or anecdotes. Lift is always shown separately from the 0-100 trust score so you can weigh both.

Question 2

Is SkillzBench available now, or is it still being built?

Accepted Answer

SkillzBench is pre-launch and actively being built by AIQ Labs. The waitlist and newsletter are live, so you can join now and be first in line when the testing lab and MCP server open up. We never claim live skill counts or published results before they exist — everything is verified before it ships.

Question 3

What kinds of agent skills will SkillzBench support at launch?

Accepted Answer

Launch scope covers instruction-only skills: SKILL.md files plus text references for agents like Claude Code, Codex, Cursor, Gemini CLI and chat-app agents. Script-bearing skills will be listed as link-outs only, so you get full verification and benchmarking where it's safe and clear pointers elsewhere.

Question 4

How does the per-agent loadout system work with my existing tools?

Accepted Answer

You connect one MCP server, then toggle skills and tools per agent to build a loadout — and each agent only sees what was chosen for it. So your Claude Code setup can differ from your Cursor setup without any cross-contamination. When an upstream skill changes, SkillzBench re-scans, re-benchmarks and notifies you.

Question 5

How is SkillzBench different from just browsing the skills repository on GitHub?

Accepted Answer

GitHub shows stars and READMEs; SkillzBench shows evidence. Every version is pinned to a commit SHA and put through validation, license checks, security scans and quality judging, then benchmarked for Lift. You also get a 0-100 trust score from security, quality, license, maintenance and community signals — and because money never touches scores, the ratings can't be bought.

Verified Agent Skills for Students