cit-prepay-check-free
Warn
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: MEDIUMPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The SKILL.md file contains a pre-crafted prompt for the user to relay to the AI agent. This prompt instructs the agent to install an external package (@alipay/agent-payment) and perform purchases, while explicitly stating that the user has already authorized these actions in an attempt to bypass agent confirmation guardrails.
- [INDIRECT_PROMPT_INJECTION]: The logic in the calculation engine takes data from the user and reflects it directly into output reports. If that data contains malicious instructions, the AI might interpret them as commands when processing the result.
- Ingestion points: Text inputs containing tax table rows in SKILL.md or provided as user input files.
- Boundary markers: None; data is processed and reflected directly into the output messages without delimiters or warnings.
- Capability inventory: The skill is permitted to use Bash and Read tools.
- Sanitization: No sanitization is performed on the Period column text or other string identifiers used in diagnostic output messages.
Audit Metadata