skill-creator

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

• Third-party content exposure detected (medium risk: 0.30). In eval-viewer/generate_review.py, the required runtime workflow embeds and renders arbitrary text from user/execution-generated files in workspace/**/outputs/* (including transcript.md/eval_metadata.json prompt text) directly into the review HTML via read_text and JSON embedding, so an outsider authoring those prompts/feedback can poison the content the UI-consumed LLM would later ingest.

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

• Potentially malicious external URL detected (high risk: 0.70). The eval viewer's HTML loads and executes remote JavaScript from https://cdn.sheetjs.com/xlsx-0.20.3/package/dist/xlsx.full.min.js at runtime (used when generate_review.py serves the viewer), which is a required external script for XLSX rendering and therefore a runtime remote code dependency.