reply-to-clara-inquiry
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted customer data retrieved from external communication logs, which could be used to influence the agent's behavior during the drafting and sending of replies.
- Ingestion points: Customer messages and inquiry details are ingested via the
get_inquirytool from the Clara platform (SKILL.md). - Boundary markers: Absent; the instructions do not specify delimiters to isolate external inquiry data from the agent's internal reasoning or drafting instructions.
- Capability inventory: The skill possesses high-impact capabilities, including sending real SMS messages via
reply_to_customerand modifying inquiry states viamark_sms_read(SKILL.md). - Sanitization: Absent; there are no instructions to sanitize or escape the content of the customer inquiries before processing.
Audit Metadata