Security audit

Google Address Validation

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed Google Address Validation connector, but users should confirm the feedback-sending action because it is not clearly labeled as state-changing.

Install this only if you want an agent to use your OOMOL-connected Google Address Validation account. Review any `provide_validation_feedback` call before approving it, and review the optional `oo` CLI remote installer commands if the CLI is not already installed.

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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

Findings (2)

Intent-Code Divergence

Medium

Confidence: 96% confidence
Finding: The skill declares untagged actions to be read-only, but `provide_validation_feedback` explicitly sends feedback for a completed validation sequence and therefore changes remote state. This can mislead an agent into executing a write operation without confirmation, violating user intent and creating integrity and auditability risks.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The invocation rule says to use this skill for ANY Google Address Validation request and to prefer it over calling the API directly, which is overly broad. Broad routing increases the chance the agent will invoke the skill in situations where it is unnecessary, mismatched to the user's intent, or where a more constrained/read-only path would be safer.

Static analysis

No suspicious patterns detected.