The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/office/soffice.py performs runtime compilation of a C shim. It writes source code to a temporary directory, invokes gcc to create a shared object, and then uses the LD_PRELOAD environment variable to inject this library into the soffice process. This mechanism is used to enable LibreOffice functionality in environments where AF_UNIX sockets are restricted.
[COMMAND_EXECUTION]: Multiple scripts invoke system commands via subprocess.run(). Specifically, scripts/recalc.py and scripts/office/soffice.py execute the soffice binary for Excel operations, and scripts/office/validators/redlining.py executes the git command to generate text diffs.
[PROMPT_INJECTION]: The skill processes user-supplied spreadsheet files, which creates an indirect prompt injection surface (Category 8).
Ingestion points: pd.read_excel in agent instructions and XML parsing in scripts/office/unpack.py and scripts/office/validators/base.py.
Boundary markers: Absent. The skill does not specify delimiters or instructions to ignore instructions embedded in cell data.
Capability inventory: Subprocess execution of soffice, gcc, and git across various utility scripts.
Sanitization: The skill uses defusedxml for XML parsing, which provides protection against XML External Entity (XXE) attacks, but does not sanitize cell content before processing.

xlsx