Security audit
Anchoring
Security checks across malware telemetry and agentic risk
Overview
This is a markdown-only coaching skill about anchoring bias, with no executable behavior, credential access, persistence, or hidden data handling.
Installers should understand that this skill may influence negotiation or financial reasoning, but it only provides user-directed coaching. Review its advice critically for your specific context, especially for high-stakes salary, valuation, legal, or business decisions.
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
63/63 vendors flagged this skill as clean.