Project Settings

GraphScribe’s options live under Project Settings → Plugins → GraphScribe.

Setting	What it does	Default
Default Format	Which format is pre-selected when you open the export dialogs. Picking a format in a dialog updates this.	Markdown
Mermaid Warn Node Count	Warns you in the export dialog when a Mermaid diagram has more than this many nodes (large diagrams get hard to read)	250
Property Dump Mode	How the reflected CDO / DataAsset property dump is collected (see below)	Modified Only
Default Save Directory	Where exported files are written by default (relative to the project directory)	Saved/GraphScribe
Open Folder After Save	Whether GraphScribe opens the output folder in your file explorer after each export	Off
Generate Graphify Viewer	After a Graphify batch export, run the graphify CLI (graphify cluster-only) so the interactive viewer (graphify-out/graph.html) and GRAPH_REPORT.md are generated. The graphify-out/graph.json wrapper is written regardless.	Off
GraphifyCliPath	Optional absolute path to the graphify CLI executable. Empty auto-detects on PATH, then the default uv install location (~/.local/bin).	(auto)

Property Dump Mode

Every export includes a reflected property dump. The actual property defaults stored on the asset (DataAsset instance or Blueprint CDO). These rows appear in the Variables sections and drive the cls: class-reference linkage. The mode controls how much inherited content that dump contains:

• Modified Only (default), properties declared on the asset’s own class plus inherited properties whose value differs from the direct parent class defaults (the editor’s definition of “modified”). Containers are compared element-wise and instanced subobjects deeply, so exports show what this class actually changed. Untouched inherited TSubclassOf / TSoftClassPtr references produce no cls: uses edges.
• Full, every Edit / BlueprintVisible property, including inherited engine properties.
• Blueprint Only, only properties declared on the asset’s own class (no inherited properties at all).

The chosen mode is stamped in JSON and Graphify output as the optional dumpMode root field (full / modifiedOnly / blueprintOnly), so consumers can tell why a dump is sparse. See Schema & Stability.

What the other settings mean in practice

• Default Format, pick Markdown if you mostly write docs, or Graphify if you mostly build AI graphs. You can always switch formats in the dialog, so this is just a convenience.
• Mermaid Warn Node Count, Mermaid is great, but a flowchart with hundreds of nodes is hard to read. If a diagram exceeds this count, the export dialog shows a warning suggesting Markdown, so you know to narrow the scope.
• Default Save Directory, set this once to a stable location (for example Documentation/GraphScribe inside your project) so exports always land in the same place. That makes it easy to diff them in git or feed them to tooling. The batch export dialog pre-fills its Output directory from this value.
• Open Folder After Save, turn this off if you export frequently and find the file explorer popping up annoying.
• Generate Graphify Viewer, a Graphify batch export always writes the machine-readable graphify-out/graph.json wrapper. This setting decides whether it also runs the graphify CLI to produce the interactive graph.html viewer and GRAPH_REPORT.md. Requires the CLI (install once with uv tool install graphifyy). A missing CLI never blocks the export. See Graphify: Install & Use.
• GraphifyCliPath, only needed when the graphify CLI is installed somewhere not on PATH (e.g. a per-project venv). Leave empty to auto-detect.

Tips

• Keep the Default Save Directory inside the repo if you want exported docs to show up in pull requests, that’s the main way teams review Blueprint changes.
• If you work with AI agents, point them at the save directory once, then re-run folder exports as your project changes. The Graphify IDs stay stable, so the graph stays useful.