Agents1Password SCAM Benchmark: Security Skills Cut Agent Failures 97%Help Net Security·high signal1200-word security skill document reduces critical credential failures from 65 to 2. Claude Opus 4.6 scores 92%, cheapest agent security measure.SourceSource pageHelp Net Security↳ Follow the threadStack layer / Threat patternMalicious agent skills evade every scanner: HKUST's SkillCloak beats detectors 90%+ of the time, SkillDetonate catches 97%Help Net SecurityPolicy dependency / Stack layerIAPS warns on Kimi Claw: security and governance risks of Chinese-hosted 'always-on' AI agentsIAPSStack layer / Threat patternArmadin details full sandbox escape in Anthropic's Claude Cowork; Anthropic disputes, silently patchesSiliconANGLEStack layer / Threat patternJim Fan: NVIDIA Open-Sources ASPIRE, and 'the Training Output Changes From Model Weights to a Continuously Expanding Skill Library'BigGo FinanceStack layer / Threat patternNSA issues official MCP hardening guidance after Amazon Q auto-execution CVEs (CVSS 8.5) enable AWS credential theftWizStack layer / Threat patternChain slash-skills: /skill-a /skill-b now loads up to 5 leading skills, not just the firstClaude Code DocsStack layer / Threat patternAnthropic ships Claude Science Workbench and an internal drug-discovery programAnthropicStack layer / Threat patternPiSAs Benchmarks Cross-User Privacy Leaks in Shared Agentic SystemsarXiv