The Agent Skills Directory

[SAFE]: The skill implements a robust, secure framework ('officecli') that forces all Microsoft Office document modifications through a temporary draft stage, requiring explicit confirmation before changes persist.
[SAFE]: Data exposure concerns are properly mitigated in the bug-reporting flow ('references/reporting.md'), which enforces a strict sanitization rule to mask document contents, values, and parameters, ensuring sensitive content remains local and confidential.
[SAFE]: Bundled external references link to well-known technology organizations (anthropics/skills) and standard open-source tools (openpyxl, pptxgenjs, pdfplumber), following a safe usage posture.
[SAFE]: User autonomy is preserved by requiring interactive mode-selection questions ('references/reviewing.md') and specific formatting constraints ('references/enduser-output.md') ensuring non-technical users have explicit oversight before changes apply.

office