The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/office/soffice.py dynamically compiles a C compatibility shim at runtime using gcc and injects it into the LibreOffice (soffice) process using the LD_PRELOAD environment variable. This technique is used to override network socket behaviors in restricted or sandboxed environments.
Evidence: The _ensure_shim function in scripts/office/soffice.py writes C source code to a temporary file and executes gcc -shared -fPIC to create a library.
[COMMAND_EXECUTION]: The skill performs various system-level operations including CLI tool execution and configuration management.
Evidence: Executes soffice, gcc, and git via subprocess.run across several scripts including recalc.py and scripts/office/validators/redlining.py.
Evidence: Automatically writes StarBasic macros to the user's local LibreOffice configuration directory (~/.config/libreoffice/... or ~/Library/Application Support/LibreOffice/...) to enable automated formula recalculation.
[PROMPT_INJECTION]: The skill processes untrusted spreadsheet data (.xlsx, .csv, .tsv) via libraries like pandas and openpyxl without utilizing structural delimiters or explicit instructions to ignore potentially malicious content embedded in the data, creating a surface for indirect prompt injection.
Ingestion points: Spreadsheet data is loaded in SKILL.md and recalc.py.
Boundary markers: None identified in the instructional content.
Capability inventory: Shell command execution (soffice, gcc, git), persistent configuration modification, and file system access.
Sanitization: No input filtering or content escaping logic was detected.

xlsx-official