Share Recording
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized commands were found in the skill's instructions or metadata.
- [PROMPT_INJECTION]: The skill processes external meeting transcripts, creating a surface for indirect prompt injection. This risk is effectively mitigated by the skill's structured extraction framework—which focuses on specific timestamps and categories—and the requirement for manual user approval (
approval_required: true) before any email is sent. - [DATA_EXFILTRATION]: The skill is designed to distribute meeting summaries to specified recipients. This data movement is the intended function of the skill and is handled via authenticated platform actions (
get_meetings,get_transcript). The logic includes specific instructions to exclude sensitive internal information when drafting emails for external recipients.
Audit Metadata