The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied markdown files as its primary input. This creates an inherent attack surface where malicious instructions embedded in a thesis could attempt to influence the agent's behavior. However, the skill implements a multi-step workflow including a diagnostic report and a mandatory user confirmation step (AskUserQuestion) before performing the polishing, which provides a human-in-the-loop checkpoint to mitigate such risks.
[COMMAND_EXECUTION]: The skill utilizes standard file system tools (Read/Write) to handle the thesis content. These operations are within the expected scope of a document processing skill and are not used to execute arbitrary shell commands or access sensitive system paths.

lovstudio-thesis-polish