Secondbrain Graph

Secondbrain Graph is a private knowledge consultation app built around a local Markdown vault. It reads the committed files in news/wiki and news/raw, parses Obsidian-style [[wikilinks]], computes backlinks, and renders the result as an interactive force-directed graph.

The project is intentionally narrow: it does not try to replicate all of Obsidian. Instead, it focuses on one job: making a growing personal news and research archive easier to explore. Notes become nodes, wiki-links become relationships, larger nodes indicate higher connectivity, and color separates concepts, sources, entities, raw articles, maps, comparisons, and meta notes. Mechanical files such as indexes and logs are hidden from the visible graph so the map stays focused on actual knowledge.

The interface combines a full-canvas graph with a compact consultation sidebar. Search surfaces matching notes with previews; clicking a node opens the note text alongside outgoing links, backlinks, source references, related notes, reading history, pinned notes, and a path finder for discovering how two ideas connect through the vault. An orbit toggle keeps isolated notes available without letting them dominate the connected core.

The app is built with Next.js, TypeScript, React Force Graph, and a Vercel-first static build flow. All parsing happens from repository files at build time, so enriching the underlying Markdown vault and redeploying refreshes the graph without changing the source folder structure or requiring uploads.