The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/office/soffice.py contains a function _ensure_shim that writes C source code to a temporary file and executes gcc to compile it into a shared object library at runtime. This dynamic code generation and compilation is a high-risk pattern.
[COMMAND_EXECUTION]: The script scripts/office/soffice.py uses the LD_PRELOAD environment variable to inject the compiled shared library into the soffice process. This technique of process interposition (hijacking system calls like socket, listen, and accept) can be used to escalate privileges or manipulate process behavior.
[COMMAND_EXECUTION]: Multiple scripts (scripts/thumbnail.py, scripts/office/soffice.py, ooxml/scripts/pack.py) use subprocess.run to execute system-level commands, including the compiler and external document processors.
[PROMPT_INJECTION]: The skill is designed to extract text and structure from untrusted user-provided .pptx and .potx files using tools like markitdown and inventory.py. This content is then ingested into the agent's context. There are no boundary markers or instructions provided to the agent to ignore or sanitize potentially malicious instructions embedded within the extracted text, creating a vulnerability to indirect prompt injection.

pptx