obsidian-layout-adjustment changes how Obsidian looks. You name a visible object such as the tabs above the note, the right sidebar or the selected icon, and the agent translates that phrase into the Obsidian layer that owns it, then edits your active CSS snippets.
Each pass follows a loop: read the active snippets from appearance.json, save a checkpoint, patch one owning layer, format the CSS, reload Obsidian and screenshot the affected area. It also covers diagnosing CSS that had no visible effect.
What You Get
- Edits to your active CSS snippets for tabs, side docks, file explorer, note surface, properties, backlinks, graph or status bar
- Named checkpoints of the snippets saved outside the snippet picker
- A screenshot of the changed area after each meaningful pass
- A short closing report of files changed, checkpoints and what was verified
Common Use Cases
- Change the tab strip color and shape
- Give the note surface a lifted look with a shadow
- Fix icons that are hard to see on a colored toolbar
- Work out why a CSS change had no visible effect
- Tidy accepted snippets into a documented file
What Makes It Unique
It is specific to Obsidian: it carries a map from everyday phrases to Obsidian layers and selector families, a list of typical mistakes such as confusing tab headers with the view header, and rules for when to restore a checkpoint. A companion reference file holds the surface map and the failure patterns.
Under The Hood
The skill is instructions for the agent plus a reference file, with no scripts. The agent reads the vault’s appearance settings and CSS snippets, copies snippets into an archive folder under .obsidian as checkpoints, and edits the active snippet files in place. It runs a CSS formatter, reloads or focuses Obsidian, and takes screenshots with the macOS screencapture tool, using AppleScript to confirm Obsidian is the frontmost window. It asks before switching themes, disabling plugins, or touching files outside the snippets and settings. It makes no web requests.