The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external Google Docs which may contain malicious instructions.
Ingestion points: The get method of the documents resource in SKILL.md allows the agent to read document body content into its context.
Boundary markers: The skill documentation does not provide specific delimiters or instructions to the agent to treat document content as untrusted or to ignore embedded instructions.
Capability inventory: The skill possesses write capabilities through the create and batchUpdate methods defined in SKILL.md, allowing it to modify data based on potentially malicious inputs.
Sanitization: There is no evidence of sanitization or validation of the content retrieved from documents before it is processed by the agent.

gws-docs