Start with local graph
Don’t stare at the full global graph trying to make sense of hundreds of nodes. Open a file you know well, right-click, and choose Open local graph. Set depth to 2 to see second-degree connections. This is far more useful for actual navigation.The graph grows organically
Everytake_notes call adds a new transcription to your memory vault, tags it, and links it to related content. You do not need to maintain the graph manually. Just keep using Augent.
Run rebuild_graph after bulk imports
If you transcribe a batch of files or change tags manually, run rebuild_graph to recompute all [[wikilinks]] and regenerate MOC files. It is idempotent and fast.
Node size reflects connections
Files with more links appear larger in the graph. MOC files and heavily-tagged transcriptions naturally become the biggest nodes, making them easy to spot as topic hubs.Multi-word tags
Tags with spaces (like “Machine Learning”) are automatically hyphenated toMachine-Learning in frontmatter. Obsidian splits bare multi-word tags into separate tags, so hyphenation keeps them as a single node in graph view.
Where your files live
- Notes (from
take_notes) save to~/Desktopby default. The augent-obsidian hard-linker moves them into your vault automatically. - Transcriptions live in
~/.augent/memory/transcriptions/and power the graph view. - Visual frames (from
visual) are saved directly into your vault’sExternal Files/visual/folder.