LLM Digest

Story

arxiv_cs_cl · Oct 26, 2026 · paper

Source brief

CoSA: The Conflict of Skills Attack in Real-World LLM Agent Ecosystems

arxiv.orgOct 26, 2026
original source linked

In brief

Agent Skills allow LLM agents to combine reusable capabilities, but checking each Skill in isolation cannot rule out policy violations when they are composed. We term such failures Conflict of Skills (CoS): artifacts...

Feed lens

agentharnesseval

Read the original at arxiv.org →Open in live feed Read that day’s brief

CoSA: The Conflict of Skills Attack in Real-World LLM Agent Ecosystems

Earlier in this thread 4 items

Can Skills Learned in Games Transfer to Real-World Work?

Show HN: shadowclone, skills learned from your coding agent corrections

SkillCorpus: Consolidating and Evaluating the Open Skill Ecosystem for Real-World LLM Agents

From Trajectories to Differentiated Skills: Cognitive Differentiation for Self-Improving Agents