analyze

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to read user-provided data files and embed "all data" verbatim into the generated dashboard HTML as inline JavaScript (const data = {...}), which forces any secrets present in those files to be included in the LLM's output.

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

• Third-party content exposure detected (high risk: 0.80). SKILL.md specifies reading arbitrary user-provided dataset files from input/<dataset-name>/ (CSV/XLS/XLSX/JSON), then writing/embedding their contents/derived text into artifacts like data-profile.md, insights.json, and dashboard.html, meaning outsider-authored free text can be ingested into the agent/LLM context via the runtime parsing and subsequent prompt/report steps.