Back to skill

Security audit

AdKit for Google Ads: Search, Performance Max, Display, Keywords and Reporting

Security checks for vulnerabilities and agentic risk

Overview

The scanners classify this skill as benign with high confidence: they report no AIG findings and one medium-severity concern that an offline conversion upload may skip the promised draft review.

The verdict is benign, but consider whether immediate offline conversion uploads fit your workflow. Review and confirm each upload carefully, since the scanner reports that this path may not use the promised draft review. The scan covered the sole reported component; no executable scripts were present.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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)

Offline conversion uploads bypass the manifest's guarantee that every change is saved as a draft.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The manifest at L007-L008 promises that every change is saved as a draft for user approval before going live. L087 instead instructs the agent to perform offline conversion uploads immediately without creating a draft. Although it still requires user confirmation, this workflow contradicts the promised draft-based review mechanism.

Content

No source excerpt is available for this finding.