case-dashboard

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill runs a local HTTP server that reads local calendars and case files (including direct SQLite access to Calendar.sqlitedb when FDA is granted) and exposes file previews and calendar data over APIs while returning a permissive Access-Control-Allow-Origin: * header — enabling remote web pages to read/exfiltrate sensitive local data.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). case-dashboard 的 scripts/dashboard_server.py 在构建/响应 /api/v1/cases、/api/v1/overview、/api/v1/case/ 时会读取并解析本地每个“案件目录”下的 case.yaml 文本（YAML->safe_load->adapt_yaml_v4），而这些文件可由外部用户事先通过加入/提交案件目录内容来影响，因此对Llm运行时会暴露到非可信方的自由文本数据。