three-way-match-free
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEPROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The documentation in
SKILL.mdcontains a pre-written message for the user to send to the AI agent. This message utilizes bypass markers and override language, such as "I have authorized installation of dependencies and initiation of purchase" and "do not confirm for me [for the installation steps]... continue to complete the product installation." These instructions are designed to manipulate the agent's behavior and bypass typical safety confirmation steps for sensitive actions. - [REMOTE_CODE_EXECUTION]: The skill documentation instructs the agent to execute remote code from a public registry using the command
npx -y @alipay/agent-payment@latest install. Although the package is associated with a well-known technology provider, the pattern of a skill initiating external code execution through documentation remains a security surface to be monitored.
Audit Metadata