The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes processes for reading content from external documents via tools like read_document and google_drive_read_file. This ingestion of untrusted data creates a surface where malicious instructions embedded within a document could potentially influence the agent's behavior.
Ingestion points: Tools read_document, google_drive_read_file, and google_drive_export_file described in SKILL.md are used to pull external document content into the agent's context.
Boundary markers: The instructions do not mention the use of delimiters or explicit warnings to the agent to ignore instructions found within the processed document text.
Capability inventory: The skill has broad capabilities including creating documents (create_document), inserting text and images, modifying styles, and performing bulk replacements or deletions.
Sanitization: There is no mention of content sanitization or validation before the agent acts on the information retrieved from the documents.

google-docs