Gen Agent Trust Hub

This skill provides a mechanism for logging agent decisions to a TSV file. It includes a security-conscious helper script that prevents spreadsheet formula injection and provides explicit warnings to agents against accessing sensitive transcripts from other projects. A minor security surface exists due to the requirement to read session transcripts for auditing, which could theoretically be used for indirect prompt injection.

The provider classified the analyzed snapshot as NONE risk. This result applies to the content available at audit time.