Verification Before Completion
Security checks across malware telemetry and agentic risk
Overview
This instruction-only skill tells the agent to verify work before claiming success and does not include hidden code, credential use, persistence, or data-sharing behavior.
Install this if you want stricter evidence-based completion reporting. Be aware that it may cause the agent to run tests, builds, linters, or similar verification commands before saying work is complete, so guide it when those commands are slow, expensive, or have side effects.
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.