The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to execute an unverified NPM package @gongrzhe/server-gmail-autoauth-mcp for OAuth authentication. This package originates from an unknown source and is not part of an established or trusted organization.
[REMOTE_CODE_EXECUTION]: The authentication step uses npx to download and execute code from the NPM registry at runtime, which is a high-risk operation when the package source is unverified.
[COMMAND_EXECUTION]: The skill requires the execution of multiple shell commands to build and restart local services (npx tsc, launchctl kickstart, systemctl restart). While these are common for development, they grant broad control over the agent environment.
[PROMPT_INJECTION]: The skill enables an Indirect Prompt Injection surface by allowing the agent to read and search external email content via the read_email and search_emails tools.
Ingestion points: Email bodies and subject lines retrieved via Gmail MCP tools (SKILL.md).
Boundary markers: None mentioned in the instructions to prevent the agent from obeying instructions embedded in email content.
Capability inventory: The agent has the ability to send_email, modify_email, create_filter, and download_attachment (SKILL.md), which could be abused if the agent is hijacked by a malicious email.
Sanitization: No sanitization or validation of the ingested email data is described.

add-gmail-agent