visual-skills

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill forces the LLM to inject user-provided dataset values directly into HTML/JavaScript (the DATA object) and output that code verbatim, so any secrets present in the uploaded or pasted data would be reproduced and exfiltrated.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill's HTML templates load and execute remote JavaScript at runtime from the ECharts CDN (https://cdnjs.cloudflare.com/ajax/libs/echarts/5.4.3/echarts.min.js), which is a required external dependency that runs code in the rendering environment.