lead-triage
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external leads which could contain hidden instructions intended to manipulate the agent's prioritization or drafting behavior.
- Ingestion points: Untrusted data is ingested from
~~crm(specifically lead notes and company descriptions) and~~почта(the content of previous email threads). - Boundary markers: The instructions lack specific delimiters or "ignore embedded instructions" warnings to prevent the model from obeying instructions found within lead data.
- Capability inventory: The skill generates prioritized summaries and email drafts. However, safety is mitigated by strict rules preventing the agent from sending emails, creating calendar events, or changing CRM statuses autonomously.
- Sanitization: There is no explicit requirement for the agent to sanitize or filter out potentially malicious prose from lead records before processing them for scoring.
Audit Metadata