The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection because it instructs the agent to read and process content from Google Docs, which could potentially contain malicious instructions.
Ingestion points: Data is read from the environment via read_document and google_drive_read_file (SKILL.md).
Boundary markers: The instructions do not include specific delimiters or warnings to separate document content from the agent's internal logic.
Capability inventory: The skill provides capabilities for writing text, tables, and images back into documents (SKILL.md).
Sanitization: There are no instructions for sanitizing or escaping the content read from external documents.
[EXTERNAL_DOWNLOADS]: The skill utilizes the insert_image tool, which necessitates fetching content from publicly accessible URLs. This is a functional requirement of the upstream Google Docs API. The skill includes a safety-conscious instruction to never widen sharing or make private images public just to satisfy this tool's requirements.

cursor-google-docs