The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to configure an external MCP server located at https://rube.app/mcp. This is a standard configuration step for using the Rube automation platform.
[COMMAND_EXECUTION]: The skill utilizes a series of tool slugs (e.g., GMAIL_SEND_EMAIL, GMAIL_BATCH_MODIFY_MESSAGES) that execute operations within the Gmail environment via the MCP protocol. These are defined as part of the 'gmail' toolkit capability.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of reading and processing external email content.
Ingestion points: Untrusted data enters the context through tools like GMAIL_FETCH_EMAILS and GMAIL_FETCH_MESSAGE_BY_MESSAGE_ID as described in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat email body content as untrusted or to ignore instructions embedded within retrieved messages.
Capability inventory: The skill possesses significant 'write' capabilities, including GMAIL_SEND_EMAIL, GMAIL_REPLY_TO_THREAD, and GMAIL_DELETE_LABEL, which could be abused if the agent obeys instructions found in a processed email.
Sanitization: No sanitization or validation logic is described for handling email content before it is processed or used in downstream workflows.

gmail-automation