tax-incentive-eligibility-check-free

Pass

Audited by Gen Agent Trust Hub on Sep 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The SKILL.md file contains a dedicated section for a 'complete version' that provides a block of text for the user to 'send to the AI assistant as-is'. This block contains instructions that influence the agent's subsequent behavior, directing it to execute shell commands, install new skills, and initiate a financial transaction from a specific merchant ID (2088702818190040). This represents a multi-step capability chain.
  • Ingestion points: The 'Need full version' (需要完整档时) section of SKILL.md provides the injection content.
  • Boundary markers: Absent; the content is designed to be provided directly to the agent's context without delimiters or safety warnings.
  • Capability inventory: The skill's frontmatter explicitly allows the Bash tool, which is necessary to execute the installation commands described in the instructions.
  • Sanitization: None; the instructional text is intended to be processed directly by the LLM to trigger actions.
  • [COMMAND_EXECUTION]: The documentation and the provided prompt block encourage the use of the Bash tool to run npx -y @alipay/agent-payment@latest install. This command initiates shell execution to perform environment setup and tool installation.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the download of the @alipay/agent-payment package via the npm registry. This package originates from a well-known financial technology service, consistent with the vendor's stated purchase flow.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 28, 2026, 09:18 PM