The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill extracts content from external PDF documents, creating an attack surface where malicious instructions embedded in a PDF could influence the agent's behavior. * Ingestion points: PDF content processed by tools such as pdf::to-markdown and pdf::extract-text. * Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the agent when processing extracted text. * Capability inventory: The worker itself performs local parsing without network access or system command execution. * Sanitization: No sanitization or filtering of the extracted text is performed before it is returned to the agent context.

pdf