The Agent Skills Directory

[COMMAND_EXECUTION]: Runtime compilation of system components. The script scripts/office/soffice.py embeds C source code which it writes to disk and compiles using gcc into a shared library. This process involves executing compiler commands at runtime.\n- [REMOTE_CODE_EXECUTION]: Process injection via shared library. The shared object compiled by the skill is loaded into the LibreOffice process using the LD_PRELOAD environment variable to intercept and wrap system calls related to network sockets.\n- [COMMAND_EXECUTION]: Dynamic macro generation and execution. The script scripts/accept_changes.py programmatically creates a LibreOffice macro and installs it into a temporary application profile to automate document editing tasks.\n- [PROMPT_INJECTION]: Ingestion of untrusted data surface.\n
Ingestion points: Extracted content from Word documents via scripts/office/unpack.py and pandoc calls.\n
Boundary markers: Absent; extracted content is interpolated into the agent context without explicit delimiters or instructions to ignore embedded commands.\n
Capability inventory: High-risk subprocess execution (soffice, gcc, git) and file system operations.\n
Sanitization: Employs defusedxml for secure XML parsing to prevent XXE attacks.\n- [EXTERNAL_DOWNLOADS]: External library dependency. The skill instructions recommend the global installation of the docx library from the public npm registry for document generation tasks.

docx