Read original ↗
paperarXivTrust 82 · PrimaryPublished 2d agoLive · 18h ago

OpenSkillRisk: Benchmarking Agent Safety When Using Real-World Risky Third-Party Skills

LLM-based agents leverage third-party skills to extend their capabilities in open-world scenarios. However, third-party skills can introduce extra security vulnerabilities, as seemingly harmless skills can contain latent safety risks that only emerge during actual execution. In this work, we conduct a systematic investigation into how well current agent systems recognize and avoid such risks. To support quantitative and qualitative evaluation, we construct OpenSkillRisk, a dedicated safety benchmark containing 263 risky skills collected from public skill marketplaces. We classify these skills

Lineage graph

Paper → model → repo connections mined from source citations (Tier-1 exact match).

Why these links exist

Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.

  • FuzzySimilar title/name (fuzzy) · 59%AgentCore-8B

    Fuzzy title match (0.73): “OpenSkillRisk: Benchmarking Agent Safety When Using Real-Wor” ≈ “AgentCore-8B”

  • PossiblePossibly related (embedding) · 53%Safety and alignment in an era of long-horizon models
  • PossiblePossibly related (embedding) · 52%What does "Safe AI" look like? [D]
  • FuzzySimilar title/name (fuzzy) · 87%SWE-agent/SWE-agent

    Fuzzy title match (0.94): “OpenSkillRisk: Benchmarking Agent Safety When Using Real-Wor” ≈ “SWE-agent/SWE-agent”

  • FuzzySimilar title/name (fuzzy) · 87%zhayujie/CowAgent

    Fuzzy title match (0.94): “OpenSkillRisk: Benchmarking Agent Safety When Using Real-Wor” ≈ “zhayujie/CowAgent”

  • FuzzySimilar title/name (fuzzy) · 66%open-multi-agent/open-multi-agent

    Fuzzy title match (0.78): “OpenSkillRisk: Benchmarking Agent Safety When Using Real-Wor” ≈ “open-multi-agent/open-multi-agent”

  • FuzzyOverlapping authors or contributors · 62%modular/modular

    Shared author/contributor keys: liu

  • FuzzySimilar title/name (fuzzy) · 59%NousResearch/hermes-agent

    Fuzzy title match (0.73): “OpenSkillRisk: Benchmarking Agent Safety When Using Real-Wor” ≈ “NousResearch/hermes-agent”

Has model

Covers

Implements (incoming)

authored (incoming)

Related across the graph

Topics