← ClaudeAtlas

research-to-obsidianlisted

Turn a linear research document (OSINT dossier, case file, investigation report, literature review, any entity-rich MD/DOCX/PDF) into an interactive Obsidian knowledge-graph vault, then install Obsidian Desktop and open it. Use when the user wants to VISUALIZE or "see the network" behind a body of research: who connects to whom, at what certainty, and why. Decomposes the material into one note per entity (person/organization/place/event/case/source), wires them with [[wikilinks]], assigns Graph-View color groups, and adds four visualizations: interactive Graph, timeline, Dataview tables, Excalidraw diagrams. Also installs Obsidian via winget, downloads the Dataview/Excalidraw/Timelines plugins, registers + launches the vault. Trigger phrases: "תהפוך את המחקר ל-Obsidian", "תבנה לי vault", "knowledge graph", "תפתח באובסידיאן", "turn my notes into a vault", "build an obsidian graph", "map the relationships". Also trigger on entity/relationship research the user wants to see visually.
hiyabh/claude-skills · ★ 0 · Data & Documents · score 72
Install: claude install-skill hiyabh/claude-skills
# Research → Obsidian Knowledge Graph `<SKILL_DIR>` = the folder that contains this SKILL.md (normally `~/.claude/skills/research-to-obsidian`). Resolve it once to an absolute path before running commands — PowerShell/cmd do not expand `~` inside quotes. The scripts below are self-contained (no dependency on files outside this folder); copy or reference them as `<SKILL_DIR>/scripts/...`. A repeatable method to convert any **entity-rich linear research** into an **interactive Obsidian vault**, then install Obsidian and open it. Proven on a multi-entity OSINT dossier (people, organizations, places, legal cases, events, sources, homonyms-to-disambiguate). **Why it works:** linear MD/DOCX hides the most valuable thing in relational research — the *network*. Obsidian's Graph View renders that network automatically from `[[wikilinks]]`, colored by entity type, so a central subject and its branches become visible at a glance, and unrelated look-alikes (homonyms) visibly disconnect. --- ## The 6-phase workflow ### Phase 1 — Read the source material Read the existing research fully (report, relations tree, sources list). Extract every **entity** and every **relationship** between entities. Note the **certainty** the research assigns to each fact (high / medium / low / confirmed-unrelated) — preserving certainty is essential for honest OSINT. ### Phase 2 — Design the entity model Classify each entity into a `type` and `role`, decide its `tags` (which drive graph color), and writ