Concept Validation

Security checks across malware telemetry and agentic risk

Overview

This is a non-executable advisory skill for planning and operating concept validation centers, with no evidence of hidden data access, persistence, or harmful behavior.

This skill is reasonable to install for concept validation center planning. Users should still treat policy, investment, IP, and compliance advice as advisory, verify current official requirements, and avoid sharing confidential project or investor details with linked skills unless those tools are trusted and appropriate.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger keywords are overly broad and include common terms such as funding, team, management, plan, and risk. In a multi-skill routing environment, this can cause the skill to activate for generic conversations outside its intended scope, leading to misrouting, irrelevant guidance, and possible unintended transfer of user context to linked skills.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal