agent-email-patterns
Pass
Audited by Gen Agent Trust Hub on Apr 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill manages inboun d email content as an ingestio n point for untruste d data in SKIL L.md (Patter n 2, 4, 6) an d reference s/multi-agent-topologies.md. Boundary marker s such as delimiter s an d instructio n framing are recommende d in reference s/security.md. The capability inventor y involve s sending email s an d creating inboxe s via the AgentMail SDK. Sanitizatio n is addresse d via the use of extracted_text to strip header s an d signature s an d explicit advice to treat email input as untruste d.
- [SAFE]: The code utilize s the agentmail librar y, which is a legitimate vendor resource associate d with the autho r agentmail-to.
Audit Metadata