The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external sources and possesses capabilities to act on that data.
Ingestion points: Tools such as gmail_messages_list, gmail_messages_get, and gmail_attachments_get (defined in SKILL.md) ingest content from external emails into the agent context.
Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" warnings for the agent to use when interpreting email body content.
Capability inventory: The skill includes powerful tools for writing and deleting data, such as gmail_messages_send, gmail_messages_trash, gmail_messages_archive, gmail_messages_move_to_label, and gmail_labels_remove (defined in SKILL.md).
Sanitization: No explicit sanitization or filtering logic is mentioned for handling external email strings before interpolation into agent prompts.
[SAFE]: The skill implements legitimate Gmail management functionality using standard integration patterns.
All described tool operations (gmail_labels_list, gmail_drafts_create, etc.) align with expected email client behavior.
Documentation correctly distinguishes between archiving and trashing, promoting best practices for data retention.
External references point to the vendor's official management infrastructure at app.hyperfx.ai.

gmail