util-0-v2-handover
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill enforces a 'memory only' constraint, instructing the agent not to read files or call tools. This minimizes the risk of unauthorized data access during the generation of the handover document.
- [PROMPT_INJECTION]: A surface for indirect prompt injection exists via the verbatim quoting of user constraints from session memory. Ingestion point: session history; Boundary markers: HOVR/2 structured format; Capability inventory: no subprocess, file-write, or network operations (explicitly disabled); Sanitization: none. This is an inherent risk of summarization tools but is mitigated by the lack of active capabilities.
Audit Metadata