The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an Indirect Prompt Injection surface by processing untrusted data from Word documents. \n- Ingestion points: The skill utilizes wps_get_active_document, getDocumentText, getComments, and getBookmarks (via wps_execute_method) to read content from the active document into the agent's context. \n- Boundary markers: The instructions do not define clear delimiters or include warnings for the AI to ignore instructions embedded within the ingested document text. \n- Capability inventory: The agent has extensive permissions to modify documents, including insertText, findReplace, setFont, and high-privilege operations like openDocument for local file access and insertHyperlink. \n- Sanitization: There is no evidence of content sanitization or validation performed on the data retrieved from the document before it is processed by the AI.

wps-word