AICP Protocol
Security checks across malware telemetry and agentic risk
Overview
This is a local Python protocol/demo skill for compact agent messages, with no evidence of hidden execution, credential access, network use, or persistence.
Safe to install as a local protocol/demo skill. If you later connect AICP messages to real agent actions, require sender validation and explicit approval for impactful operations such as create, update, delete, or query.
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
67/67 vendors flagged this skill as clean.