email-response-simulation
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is primarily instructional and provides a structured workflow for text analysis and scoring. No malicious executables, hidden commands, or dangerous network operations were found.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of email drafts and enrichment data which could contain embedded instructions intended to influence the agent.
- Ingestion points: The skill ingests
Email to reviewandAccount enrichment JSONas described in the Inputs section ofSKILL.md. - Boundary markers: There are no explicit delimiters or specific instructions for the agent to ignore instructions embedded within the email drafts being reviewed.
- Capability inventory: The skill's capabilities are limited to reading local campaign context and writing review results to a specific directory (
campaigns/{campaign-slug}/reviews/). It does not demonstrate shell access or network exfiltration capabilities. - Sanitization: The skill does not define any sanitization or filtering steps for the ingested email content.
Audit Metadata