find-orders
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides explicit instructions to the agent to default to a read-only state. It contains safeguards requiring the user to confirm any account-changing actions such as cancellations or returns, which limits the risk of unintended consequences during browser automation.
- [DATA_EXPOSURE]: The skill is designed to access personal information, including Amazon purchase history, delivery status, and invoices. This access is limited to the user's active browser session and aligns with the stated primary purpose of the skill.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external web pages (Amazon.com). This presents a standard surface for indirect prompt injection where malicious instructions could be embedded in product titles or descriptions. However, the instructions direct the agent to capture and summarize only specific factual fields (dates, totals, status), which helps mitigate the risk of the agent following instructions embedded in that data.
- Ingestion points: Amazon order history and detail pages accessed via browser automation.
- Boundary markers: None explicitly defined for the external data.
- Capability inventory: Browser automation and data summarization.
- Sanitization: Not explicitly defined.
Audit Metadata