The Agent Skills Directory

[SAFE]: The skill implements robust security pre-checks on documents to prevent decompression attacks (zip bombs). It strictly limits the number of entries in a package, the uncompressed size of individual entries, and the total uncompressed size of the document before processing.
[SAFE]: A core security invariant of the skill is that source documents are treated as read-only. The scripts are designed to create new files and explicitly refuse to overwrite existing files or modify the original source path.
[SAFE]: All file operations are restricted to absolute paths, and the skill validates that output paths do not resolve to the source file, preventing accidental data loss or corruption.
[SAFE]: The skill uses established and well-known document processing libraries for its operations, including openpyxl, python-docx, pypdf, and python-pptx.
[SAFE]: Editing operations are designed to be conservative; the tool refuses to perform edits on complex document structures (such as shared formulas in spreadsheets or bookmarks in Word documents) that could lead to unintended data loss or broken document integrity.
[SAFE]: The skill instructions provide the agent with verification steps to confirm the integrity of the generated files using standard readers before reporting success, ensuring the output is valid and safe to open.

office-transform