The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data and files which may contain malicious instructions.
Ingestion points: The generate_from_template function in SKILL.md ingests content via the template_path and content dictionary.
Boundary markers: The implementation lacks explicit delimiters or instructions to the agent to ignore embedded commands within the processed files.
Capability inventory: The skill includes functions that call subprocess.run to execute system utilities like zip, unzip, and soffice (SKILL.md).
Sanitization: There is no evidence of input validation or sanitization before the data is interpolated into the presentation's content.xml file.
[COMMAND_EXECUTION]: The skill relies on executing system-level commands through Python's subprocess.run module to handle presentation file manipulation (zipping/unzipping) and conversion via the soffice binary. While the use of argument lists mitigates shell injection, this remains a significant capability that interacts directly with the host environment.

Impress