Create New Skill

Security checks across malware telemetry and agentic risk

Overview

This skill is a guided assistant for creating other skills, and its persistent file-writing behavior is disclosed and confirmation-gated.

Before installing, prefer workspace scope unless you intentionally want the new skill available across all projects. Review the generated trigger conditions, tools, workflow, constraints, and final file preview before confirming writes, especially when the skill is generated from external or local documentation.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Vague Triggers

Medium
Confidence
84% confidence
Finding
The example trigger phrase "Help me create a skill" is broad, natural language that could easily appear in ordinary conversation and may cause the skill to activate when the user did not explicitly intend to enter a skill-creation workflow. In this context, unintended activation is meaningful because the skill guides creation of new agent behaviors, which can lead to prompt-surface expansion, confusion, or accidental generation of unsafe or overprivileged skills.

Self-Modification

High
Category
Rogue Agent
Content
If confirmed:
- Create the skill directory.
- Write SKILL.md with full behavioral instructions.
- Write README.md.
- If directory exists, stop and ask.
Confidence
95% confidence
Finding
Write SKILL

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal