ColluSkill splits a malicious intent across three interdependent agent skills to evade six skill scanners in 96.0% of cases, while ChainGuard reduces this to 22.5% by analyzing a candidate skill with its installed-skill context.
https://github.com/cisco-ai-defense/skill- scanner
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
ColluSkill: Adversarial Cross-Skill Composition for Evading Agent Skill Scanners
ColluSkill splits a malicious intent across three interdependent agent skills to evade six skill scanners in 96.0% of cases, while ChainGuard reduces this to 22.5% by analyzing a candidate skill with its installed-skill context.