AI Automation Agency Blueprint
Security checks across malware telemetry and agentic risk
Overview
This is a documentation-only business playbook for AI automation agencies, with no code or hidden system access requested.
Safe to install as an advisory business skill. Treat its pricing, legal, insurance, and scaling recommendations as general guidance, verify them for your jurisdiction and clients, and review optional paid context-pack links before purchasing.
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
65/65 vendors flagged this skill as clean.