chart-visualization
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use the
curltool to perform API requests for generating charts. - [EXTERNAL_DOWNLOADS]: The skill communicates with an external API at
https://antv-studio.alipay.com/api/gpt-vis. This endpoint is a public service provided by AntV (part of Ant Group/Alipay), which is a well-known service for data visualization. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data to construct JSON payloads for API requests. This creates an attack surface where maliciously crafted user data could attempt to influence the agent's behavior or the parameters of the tool call.
- Ingestion points: User input data is used to populate fields like
data,title, and categories in the chart request (SKILL.md). - Boundary markers: The instructions do not define clear delimiters or provide the agent with explicit warnings to treat the user data as untrusted content.
- Capability inventory: The agent uses the
curltool to send data to a remote service and processes the resulting URL to generate a markdown image link (SKILL.md). - Sanitization: The skill lacks explicit instructions for the agent to validate, sanitize, or escape the user-provided data before it is interpolated into the command payload.
Audit Metadata