results-reporting

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to "include exact artifact paths and commands supplied by the user" and to allow reproduction metadata to "contain whatever the user actually has," which forces verbatim inclusion of user-supplied commands — and any secrets embedded in them — in the agent's output.

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/render_results_report.py the runtime reads a user-supplied JSON artifact from --json, and the renderer includes fields like question, data, methods, results, validation, interpretation, limits, and reproduction verbatim into the produced Markdown, so outsider-authored free text can be ingested without selecting a specific item first.