Security audit

Adagent Google Ads

Security checks for vulnerabilities and agentic risk

Overview

This Google Ads connector is coherent, but it uses a password-like MCP link with authority to read and change ad accounts, so it needs careful review before installation.

Install only if you trust AdAgent with your Google Ads account. Treat the MCP link as a secret, verify the Google OAuth permissions, confirm account, budget, geography, language, and final URL before any campaign creation or enablement, and revoke or regenerate access immediately if the link is exposed.

A.I.G

Vulnerability Patterns

Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
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

Findings (1)

T09 · Insecure Skill Coding Practices

Error

Location: SKILL.md:14
Finding: Credential-Bearing MCP URL Is Vulnerable to Disclosure and Replay
Content: View full analysis
Remediation: View remediation

SkillSpector

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 (3)

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 92% confidence
Finding: The trigger description includes broad phrases like 'keyword research', 'search ads', and 'ad performance', which can match routine informational or marketing queries without clear user intent to manage a Google Ads account. In a skill that can read account data and create or modify campaigns through an authenticated MCP link, overbroad activation increases the risk of unintended tool use and exposure of sensitive account information or state-changing actions.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 86% confidence
Finding: The keyword research example hard-codes Taiwan and Chinese without indicating that these are merely examples or requiring user choice. This can steer the agent toward processing requests in an unintended market or language context, producing misleading ad research and potentially causing privacy or business-context errors if applied automatically to a user's account workflow.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 89% confidence
Finding: The campaign creation example hard-codes a specific location and language, which may cause a campaign to be prepared for the wrong market if the agent follows the example by default. In this skill context, even though campaigns start paused, generating campaigns with incorrect targeting can still lead to operational mistakes and accidental activation of misconfigured ads.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.