research-report

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

• Third-party content exposure detected (high risk: 0.75). The skill reads output_dir/*.json produced by /research-deep (which can include outsider-authored web content or other third-party text) and then executes a generated script that ingests those JSON values as readable strings into {topic}/report.md via the LLM-driven report generation flow.