Skip to main content
Every transcription, note, and translation Augent creates is a markdown file with structured YAML frontmatter, tags, timestamps, source URLs, and [[wikilinks]] between related content. Point Obsidian at the memory directory and you get a live knowledge graph that grows with every take_notes call. Augent knowledge graph in Obsidian

Obsidian Setup for Mac

Set Obsidian as the default opener for .md and .txt files on macOS.

​
What you get

  • YAML frontmatter on every file: title, tags, duration, language, date, type
  • Automatic tagging: semantic embeddings assign topic tags without any manual work
  • [[Wikilinks]] between semantically related transcriptions
  • MOC (Map of Content) hub files that cluster transcriptions by topic
  • Notes ↔ transcription links: every notes file links back to its source
  • No orphan nodes: chronological neighbor links ensure every file is connected

​
File types

​
How connections form

  1. Tags create shared hub nodes. Two files tagged #AI both connect to the AI node
  2. [[Wikilinks]] connect semantically similar transcriptions via embedding cosine similarity
  3. MOC files act as cluster centers, pulling related nodes together
  4. Notes link back to their source transcription via source_transcription frontmatter
  5. Translations link back to the original via original frontmatter
  6. Chronological neighbors: every new transcription links to the previous one, preventing orphans

​
Graph view

​
Color groups

Open Graph view > Settings (the gear icon) and add color groups to visually distinguish topics. These are examples to get you started — replace them with your own tags and colors: Add more groups as your tags grow. Any query that works in Obsidian search works here. Obsidian graph with color-coded topic clusters

​
Display settings

These values are a good starting point. Tweak them to your preference. Enable Arrows to see link directionality. Turn Tags off in the filter section — at scale, tag nodes add clutter without adding information.

​
Local graph vs global graph

  • Global graph (Cmd + G): Every file in the vault. Great for the full picture, but gets dense past a few hundred files.
  • Local graph: Right-click any file > Open local graph. Shows only that file’s immediate connections.
Local graph is where graph view becomes genuinely useful. Set depth to 2 to see second-degree connections. Graph view showing topic clusters

​
Filtering

In the graph filter section, type tag:#AI to show only AI-tagged files and their connections. Combine with -path:MOC to hide hub files and see direct file-to-file links.