Security audit

Google Workspace Cli

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed Google Workspace automation guide, but it encourages non-interactive actions that can send email and change live business data without strong user confirmation safeguards.

Install only if you intend to let the agent operate a real Google Workspace account through the gog CLI. Use least-privilege OAuth scopes and test accounts where possible, keep --no-input disabled for risky tasks, and require explicit review before sending mail, modifying Sheets, deleting calendar data, or exporting sensitive documents.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Intent-Code Divergence

Medium

Confidence: 90% confidence
Finding: The skill’s security section claims command execution is limited to whitelisted commands and avoids user-input command construction, but the document elsewhere provides shell and Python automation examples that pass variable data into command invocations. This mismatch can mislead users and agent implementers into overtrusting the skill, increasing the chance that unreviewed user-controlled values are used in sensitive exec flows.

Vague Triggers

Medium

Confidence: 87% confidence
Finding: The skill advertises very broad activation conditions such as improving efficiency, automation, workflow optimization, and general productivity use. In an agent environment, this can cause the skill to trigger for many ordinary requests and perform high-impact external actions like sending mail, modifying documents, or accessing accounts without sufficiently narrow user intent confirmation.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The batch email example demonstrates automated outbound transmission to many recipients without a prominent warning that this sends real external communications. In an agent setting, users may treat the example as harmless automation and accidentally trigger mass mail, privacy exposure, or phishing-like behavior.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill documents automated Sheets write, append, and clear operations without a strong warning that these modify live spreadsheet data. In production or enterprise contexts, this can lead to accidental corruption, destructive changes, or unauthorized updates if the agent over-applies the skill to ambiguous requests.

Missing User Warnings

Low

Confidence: 81% confidence
Finding: The batch export example writes documents to local files without warning about local persistence, possible overwrites, and storage of sensitive content outside Google Workspace controls. This can expose confidential documents to less protected disk locations or leave residual copies in shared environments.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.