Ad Ops & Cross-Channel Advertising Agent

Security checks across malware telemetry and agentic risk

Overview

This looks like an ad-operations playbook, but it gives autonomous budget and campaign-optimization guidance without clear approval, spend-limit, or rollback controls.

Review before installing if the agent has or may receive access to Google Ads, Meta, LinkedIn, TikTok, or other ad accounts. Treat it as advisory unless you add strict controls: read-only by default, explicit approval for campaign/bid/budget/audience/creative changes, spend-change limits, audit logs, and rollback steps.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill explicitly frames itself as an autonomous ad-operations manager that can plan, audit, optimize, and report across paid media channels, but it does not state any guardrails requiring read-only mode, human approval, budget caps, or sandboxing before changes are made. In this context, optimization actions can directly alter bids, targeting, pacing, and spend on live campaigns, creating a real risk of unintended financial loss or business disruption if an agent is granted connected platform access.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal