SkillTrustBench
Safety & securityVerifiedLast updated 2026-10-04
Security benchmark for agent skills, distilled from real skill marketplaces.
SkillTrustBench, released in June 2026 by Tencent Zhuque Lab and CUHK-Shenzhen, distilled 5,520 evaluation cases covering nine attack categories from 62,652 real skills found in mainstream skill marketplaces.
Key facts
| 2026 frontier | 5,520 evaluation cases across 9 attack categories (T01-T09) and 5 dependency layers |
|---|---|
| Credibility | Notable for scale of provenance: the cases were distilled from 62,652 real skills taken from major skill marketplaces, so it tests the threats that actually ship rather than hand-invented ones. Published June 2026 by Tencent Zhuque Lab with CUHK-Shenzhen. |
| Verification | Verified |
| Source | Tencent Zhuque Lab / CUHK-Shenzhen |
FAQ
What does SkillTrustBench measure?
Security benchmark for agent skills, distilled from real skill marketplaces.
What is the 2026 frontier for SkillTrustBench?
5,520 evaluation cases across 9 attack categories (T01-T09) and 5 dependency layers
Is SkillTrustBench credible?
Notable for scale of provenance: the cases were distilled from 62,652 real skills taken from major skill marketplaces, so it tests the threats that actually ship rather than hand-invented ones. Published June 2026 by Tencent Zhuque Lab with CUHK-Shenzhen.