What problem does it solve? After a deep research run collects raw evidence (evidence.jsonl, sources), turning that scattered material into a decision-ready report is slow and error-prone. This Skill validates the research run, groups claims with their supporting evidence, surfaces contradictions, and produces a structured report plus an optional implementation plan. ## Core Features & Use Cases - Evidence Verification & Grouping: Reads input.yaml and evidence.jsonl, links every claim to evidence IDs, flags low-confidence items (score <= 2), and records contradictions and unresolved questions. - Structured Report Generation: Produces report.md with an executive summary, key findings with confidence levels, an evidence table, contradictions, predictions, recommendations, and next steps. - Implementation Planning: With focus=implementation, generates implementation_plan.md including architecture, data model, phased task breakdown, test plan, and operations design. - Use Case: After running dr-explore to gather sources on a technical topic, invoke this Skill on the run directory to get a decision-ready report with every claim traceable to a cited source. ## Quick Start Synthesize the latest deep research run into a report with focus on implementation planning for an internal audience.
