What problem does it solve? Turning verified OSINT findings into a publishable report risks citation hallucination, inconsistent formatting, and unverifiable claims. This Skill separates editorial synthesis (done by the model) from deterministic rendering and validation (done by code), so every citation, verdict, and confidence level traces back to ground-truth case files. ## Core Features & Use Cases - Structured draft authoring: The model writes report-draft.json with title, deck, finding order, headlines, summaries, caveats, and next steps, referencing only fact-checked finding IDs. - Deterministic rendering: Code validates finding references and produces three artifacts: findings-report.md (claim-by-claim audit), report.html (designed publication artifact), and evidence-map.json (machine-readable claim-to-source ledger). - Diagram selection: Optionally select flow, hierarchy, network, loop, timeline, or bar diagrams from existing findings data; the renderer owns all Mermaid, SVG, and layout output. - Use Case: After an OSINT investigation passes its fact-check gate, use this Skill to synthesize the verified findings into a publication-ready HTML report with an auditable evidence ledger, without ever hand-editing generated HTML. ## Quick Start Ask the agent to run Phase 5 report drafting on the current case directory so it authors report-draft.json from findings.json and fact-check.json, then renders the final Markdown, HTML, and evidence map.
