The Agent Skills Directory

[DYNAMIC_EXECUTION]: Library injection and runtime compilation. The script 'scripts/office/soffice.py' writes a C source string to a temporary location and compiles it into a shared library using 'gcc'. It then utilizes the LD_PRELOAD environment variable to inject this library into the 'soffice' process to shim system calls.
[DYNAMIC_EXECUTION]: Macro generation. 'scripts/recalc.py' generates a LibreOffice StarBasic macro and writes it to the local environment to enable automated calculation and saving of workbooks.
[PERSISTENCE]: Configuration modification. 'scripts/recalc.py' writes a macro file to the user's LibreOffice configuration directory (e.g., ~/.config/libreoffice/ or AppData/Roaming/LibreOffice/), which remains present across application sessions.
[COMMAND_EXECUTION]: System utility invocation. The skill uses 'subprocess.run' to execute 'soffice', 'gcc', 'git', and 'timeout' across various scripts to process documents and manage environment shims.
[INDIRECT_PROMPT_INJECTION]: Vulnerability surface via data ingestion. (1) Ingestion points: The skill reads and unpacks Office XML content from potentially untrusted files in 'scripts/office/unpack.py' and 'scripts/recalc.py'. (2) Boundary markers: Absent; there are no specific instructions or delimiters to isolate data from commands. (3) Capability inventory: The skill has broad execution capabilities including subprocess calls to compilers and office suites. (4) Sanitization: The skill uses 'defusedxml' and 'lxml' for XML parsing, providing protection against typical XML exploits like Billion Laughs.

xlsx