The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill performs runtime compilation and process injection to ensure tool compatibility in restricted environments.
scripts/office/soffice.py writes C source code to a temporary file and compiles it into a shared object using gcc.
The resulting library is injected into the LibreOffice (soffice) process via the LD_PRELOAD environment variable to modify low-level socket handling behavior.
[COMMAND_EXECUTION]: The skill invokes multiple system utilities through the Python subprocess module for document processing tasks.
scripts/recalc.py executes the soffice binary to update spreadsheet formula values and utilizes timeout or gtimeout for process lifecycle management.
scripts/office/soffice.py calls gcc to compile its compatibility shim at runtime.
scripts/office/validators/redlining.py uses git diff to perform word-level comparisons of document versions.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its data ingestion capabilities.
Ingestion points: Reads content from external spreadsheet and Word document files (.xlsx, .csv, .docx) in scripts/recalc.py and via instructions in SKILL.md.
Boundary markers: Lacks explicit delimiters or instructions to ignore instructions embedded within the user-provided data files.
Capability inventory: Provides significant file system access and system command execution capabilities through the document processing toolchain.
Sanitization: Employs defusedxml to mitigate XML External Entity (XXE) attacks, but does not sanitize the textual or formula content against logic-based injection.

xlsx