The Agent Skills Directory

[SAFE]: The skill operates as a local document utility for editing OOXML structures (.docx). Its core logic in docxmod_skill.py utilizes the standard zipfile and lxml libraries to perform targeted edits on internal XML parts. No network connections, credential harvesting, or data exfiltration mechanisms were found.
[SAFE]: The skill includes a comprehensive self-contained test suite (tests/test_docxmod.py) that validates the 'lossless' claim by verifying that unmodified document parts remain binary-identical after a save operation. This transparency is a positive security indicator.
[SAFE]: Dependencies are restricted to well-known and reputable packages (lxml, PySide6, pytest). The skill does not download or execute remote scripts at runtime.
[SAFE]: There are no signs of prompt injection or deceptive instructions. The command structure and workflow are clearly documented and correspond directly to the provided code functionality.
[SAFE]: File operations are handled safely, with in-place saves implemented via temporary files and atomic moves to prevent data corruption. Output defaults to a new file to protect original user data.

docx-editor