Self Improving
Security checks across malware telemetry and agentic risk
Overview
This markdown-only skill transparently keeps local correction notes to tune future agent behavior and does not show hidden code, credential use, or network activity.
Install only if you want the agent to keep local notes about corrections and use them to adjust future behavior. Keep autonomous mode off unless you deliberately want less-interrupted behavior tuning, and periodically review or clear memory.md, corrections.md, and experiments.md if the agent starts following an unwanted preference.
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.