The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands such as unzip and rm -rf using variable placeholders like {파일} and {임시디렉토리}. This creates a command injection surface if the agent populates these variables with unsanitized file paths containing shell metacharacters.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of the pyhwpx library via pip. As an external dependency not part of a standard trusted environment, it should be verified before installation.
[PROMPT_INJECTION]: The skill involves processing content from external HWPX files. 1. Ingestion points: The agent reads text from Contents/section0.xml. 2. Boundary markers: None are defined to separate document content from agent instructions. 3. Capability inventory: The agent has access to shell execution and Python environments. 4. Sanitization: There is no logic to sanitize or escape document text. This creates a risk of indirect prompt injection where instructions hidden in the HWPX document could influence agent behavior.

hwp-analyze