{"skill":{"slug":"deep-research-v22","displayName":"Deep Research Pro v2.2","summary":"Conducts thorough research by downloading full PDFs, extracting structured data with original text quotes, verifying sources, and generating cross-validated...","tags":{"deep":"2.2.0","latest":"2.2.0","pdf-extraction":"2.2.0","research":"2.2.0","study":"2.2.0"},"stats":{"comments":0,"downloads":143,"installsAllTime":0,"installsCurrent":0,"stars":0,"versions":1},"createdAt":1773917853115,"updatedAt":1773918418828},"latestVersion":{"version":"2.2.0","createdAt":1773917853115,"changelog":"Deep Research Pro v2.2.0 introduces strict full-text extraction, structured data requirements, and enforced source verification:\n\n- Mandatory full-text parsing and structured data extraction for all sources; sources without full text are excluded from conclusions.\n- New planning step: users must confirm at least 5 concrete search queries (`research/plan.md`) before proceeding.\n- Only sources with confirmed full-text extraction (≥10) count towards results; each must include copied quotes, concrete data, and scoring.\n- Automatic provenance check script is now required before report generation; reports with untraceable data are blocked.\n- Enhanced cross-analysis: find and document at least 3 sets of conflicting data, clearly citing source cards.\n- Strengthened quality gates: rejects incomplete extractions, missing provenance, or insufficient card count.","license":"MIT-0"},"metadata":null,"owner":{"handle":"xueylee-dotcom","userId":"s17cfbpghyxeebsx1t64sx4ck583q0e9","displayName":"xueylee-dotcom","image":"https://avatars.githubusercontent.com/u/260164558?v=4"},"moderation":null}