Agentic Loop Designer

Security checks across malware telemetry and agentic risk

Overview

This is a documentation-only skill for designing automation loops, with disclosed and mostly purpose-aligned risks users should configure carefully.

Before installing, review each loop for data sensitivity, write access, and destinations. Require human approval for external messages, confidential data, account or business-record changes, and irreversible actions; replace silent discard rules with a visible review queue or audit trail if submitters expect their ideas to be considered.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
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

Findings (1)

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The skill explicitly recommends silently discarding Slack messages tagged with #idea when the model score is below 50. That creates a hidden automated moderation path over user-submitted content without notice, appeal, or auditability, which can cause data loss, unfair filtering, and user distrust if the model misclassifies legitimate ideas. In this skill's context, the danger is increased because the workflow is presented as a ready-to-deploy template and encourages autonomous operation.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal