Synthesize
Security checks across malware telemetry and agentic risk
Overview
This is a documentation-only skill that helps an agent combine sources and handle conflicting information, with no evidence of hidden execution or sensitive access.
Installing this skill should mainly change how the agent organizes research and reports uncertainty. Review its preferred output style and conflict-resolution rules if those matter for your workflow, but there is no artifact evidence of hidden execution, credential use, persistence, or sensitive local access.
SkillSpector
By NVIDIA
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
VirusTotal
66/66 vendors flagged this skill as clean.