AgentPod is building a private, secure device for your AI agent.The AgentPod device is coming.Coming soonBe first
We checked this and rejected itsecurity

mcp-google-workspace: why it is not safe to use

We found broad write actions in Google Workspace, including delete, archive, and send operations, without an in-skill confirmation step.

What we found

Our review did not find hidden prompts, hardcoded secrets, obfuscated code, undisclosed remote fetches, or access to unrelated credential stores. Network use appeared limited to Google OAuth and Google Workspace APIs that the user chooses to authorize.

The main concern is operational safety. This skill can take irreversible or hard-to-undo actions in Gmail and Google Calendar when the relevant permissions are enabled. Based on the repo, those actions can include deleting calendar events and Gmail drafts, archiving messages, and sending replies. We did not see an in-skill confirmation step before those actions occur. That means a mistaken instruction, overly broad prompt, or automation error could cause changes in a live mailbox or calendar without a final check inside the skill.

What to do instead

Use this skill only with the minimum Google scopes needed, and prefer read-only access where possible. If write access is required, use it in workflows that add an explicit human approval step before delete, archive, or send actions. For higher-risk environments, test against a secondary Google account or non-production calendar first, and limit which accounts and data the skill can access.

Want the same outcome, safely? Use our checked skill instead.

Source: https://github.com/j3k0/mcp-google-workspace

We report what our security review found at the time we checked, with the goal of keeping people safe. Projects change; if a maintainer has since fixed this, we are glad to recheck it. Email hello@agentpod.com.

Copied to clipboard. Paste it into your AI (ChatGPT, Claude, or your agent) to add the skill.