recoup-internal-consulting-progress-report
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE]: The skill is designed to read sensitive project information, including executed agreements, email correspondence, and meeting transcripts. However, it implements strict boundaries, such as restricting access to the current project's workspace, prohibiting searches of other private checkouts, and requiring that all qualitative claims be attributed and labeled as estimates. Access to this data is necessary for the skill's primary function of internal reporting.
- [DATA_EXFILTRATION]: While the skill prepares delivery formats (HTML, Slack, or email), it explicitly states that the agent should default to a draft and only send the content when the user has authorized the specific recipient and content. This human-in-the-loop requirement mitigates the risk of unauthorized data exfiltration.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as meeting transcripts and emails. It mitigates potential injection risks through a multi-step review process, including a 'copy-reviewer' pass and a 'fresh-context condense pass' intended to identify inconsistencies or inaccuracies before finalization.
Audit Metadata