The Agent Skills Directory

[SAFE]: The skill follows a benign instructional design focused on document production and refinement.
[SAFE]: Access to external data sources such as Slack or Google Drive is mediated through user-approved integrations and serves the primary goal of context gathering for writing tasks.
[SAFE]: Document manipulation is performed using standard agentic tools like create_file and str_replace to update local artifacts, with no evidence of malicious command execution or data exfiltration.
[SAFE]: The reader testing phase uses isolated sub-agents or manual user testing to verify document quality, which is a security-positive pattern for validating content without context bleed.

doc-coauthoring