The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes command-line tools provided by the jinko Python package, such as jinko.cli.create_document_from_markdown, to facilitate document creation and updates. These tools are part of the intended vendor-provided SDK surface.
[INDIRECT_PROMPT_INJECTION]: Because the skill processes external markdown content, there is a potential surface for indirect prompt injection. The skill mitigates this risk with explicit rules instructing the agent to treat all markdown as data rather than instructions and includes dedicated evaluation cases to verify this behavior.
[DATA_EXPOSURE]: The skill manages session-specific identifiers and credentials. It implements safety mechanisms to prevent data exposure, such as enforcing directory containment for image uploads and explicitly instructing the agent to reject attempts to access sensitive files like .env through symlinks or path manipulation.

jinko-document