Output

An export writes up to four formats into one output folder: an HTML site, Markdown files, a JSON report and AI-ready text. You choose the formats on the Export page of the panel, in the settings, with -Formats on the command line or in the export options from Python.

A live sample shows a generated HTML site.

Output folder

The default output folder is Saved/BlueprintScribe in your project folder. Relative paths start at the project folder; absolute paths are used as they are. With all formats on, the folder looks like this:

Saved/BlueprintScribe/
  .blueprint-scribe     marker file, see "Cleaning" below
  index.html            start page of the HTML site
  audit.html            all findings (only with Audit findings on)
  migration.html        C++ migration report
  assets/               scribe.css, scribe.js, data.js and icon.svg of the site
  pages/                one .html page per asset
  images/               graph images, one folder per asset
  markdown/             index.md, audit.md, migration.md and one .md file per asset
  scribe.json           JSON report
  ai/                   AI-ready text: one .md per asset, blueprints.md and blueprints.jsonl

Every asset has an id that is used for its file names: the package path without the leading slash, with / replaced by .. Characters other than letters, digits, ., _ and - become _. For example, /Game/Blueprints/BP_Door is written to pages/Game.Blueprints.BP_Door.html, markdown/Game.Blueprints.BP_Door.md, ai/Game.Blueprints.BP_Door.md and images/Game.Blueprints.BP_Door/.

Tip

Saved is usually not under source control. To commit the documentation or publish it, set the output folder to something like Docs/Blueprints.

Cleaning

Every export writes a .blueprint-scribe marker file into the output folder. Before an export, Blueprint Scribe checks for this file:

• Marker present (an earlier export wrote to this folder): the pages, images, markdown, assets and ai folders and the files index.html, audit.html, migration.html and scribe.json are deleted, then the new export is written. Nothing else in the folder is touched.
• No marker: nothing is deleted. Files with the same names are overwritten.

This keeps pages of deleted or renamed Blueprints from lingering, without ever deleting a folder that Blueprint Scribe didn’t create.

Caution

After the first export, the folder has the marker. Don’t keep your own files in its pages, images, markdown, assets or ai folders: the next export deletes them. Also note that an export of a single asset (Export page, Export Blueprint Docs) into the same folder replaces a previous export of the whole project.

HTML site

Open index.html in a browser. The site needs no server: the navigation and search data are loaded from assets/data.js as a script, so the site works from a local folder or an unpacked zip, and you can upload the folder to any static web host. It loads no files from the internet.

Layout

Every page has the same frame:

• Top bar: the title, Overview, Audit and C++ migration links (Audit only when audit findings are included), the search box and a button that switches between light and dark theme.
• Sidebar: every folder as a collapsible group with its assets sorted by name, a colored dot for the asset kind and a marker on assets with errors or warnings. The Filter assets box filters the list by name. On narrow screens the sidebar opens with the menu button.
• Footer: the export time and the credit line.

Search

The search box (Search Blueprints, functions, variables) searches asset names, descriptions and folders, and the names of variables, functions, events, dispatchers and macros. All words you type must match, ignoring case. Exact names come first, then names that start with the query, then names that contain it, then other matches (for example in a description); at the same level, assets come before members. Up to 40 results are shown.

Key	Action
/	Jump to the search box (when you are not typing in another field)
Arrow Down / Arrow Up	Move through the results
Enter	Open the highlighted result, or the first one
Escape	Close the results

Themes

The site follows the light or dark setting of your operating system. The theme button switches between light and dark, and the browser remembers your choice.

Overview page

index.html shows the title, the scanned scope and the export time, and these numbers: Blueprints, Structures and enums, Nodes, Documented, Complexity (average), and with audit findings included Health score, Errors and Warnings.

Below is a table of all assets with Name (and the start of the description), Kind, Folder, Nodes, Complexity, C++ (the C++ migration score), Documented and, with audit findings included, Issues (errors and warnings). Assets without logic show - for Complexity and C++. Kind chips filter the table, and clicking a column header sorts it.

Audit page

audit.html lists every finding with Severity, Asset, Rule and Finding, errors first. Chips filter by severity and by rule. The asset name links to the audit section of the asset’s page. The page is written only when Audit findings is on.

C++ migration page

migration.html shows the number of strong C++ candidates, of Blueprints to consider, of Blueprints with upgrade notes and the average complexity, then a sortable table with Asset, C++ score, Verdict, Complexity and Why (the reasons and the upgrade notes), highest score first. The asset name links to the C++ migration section of its page. See C++ migration report.

Asset pages

Each asset page starts with the folder, the kind and the name, the description (or “No description.”), and a list of facts: Parent, Interfaces, Display name, Category, Namespace, Skeleton, Compile, Size, Complexity and C++ candidate (for assets with logic), Hard references, Used by and Documented. Facts that are empty are left out.

The sections below depend on the asset. An On this page list next to the content links to each section.

Section	Content	Assets
Audit	The asset’s findings with severity, rule and message, linked to the graph they are in	Any, with audit findings included
C++ migration	The C++ score and verdict, the reasons, and Before an engine upgrade with the upgrade notes	Assets with reasons or upgrade notes
Entries	Name, value and description of each entry	Enums
Components	Component tree with classes, inherited components marked “from” the class that adds them, and tooltips	Actor Blueprints
Widgets	Widget hierarchy, widgets marked Variable when they are exposed as variables, and widget animations	Widget Blueprints
Variables / Fields	Name with flags, Unused marker and category; type; default value; description	Blueprints / structures
Functions	One card per function, see below	Blueprints with functions
Events	One card per event, with a link to the graph that holds it	Blueprints with events
Event Dispatchers	One card per dispatcher with its signature	Blueprints with dispatchers
Macros	One card per macro	Blueprints with macros
Graphs	Table of all graphs with kind and node count, and the images of graphs not already shown with a function or macro	Blueprints
Notes	Comment boxes and node comments, with their graph and node	Blueprints with comments
References	Uses (hard references, then soft ones marked “soft”) and Used by (levels of World Partition actors marked “placed in level”)	Any with references

A function card shows the name, the access when it isn’t Public, the flags, the signature (for example Open(Speed: Float (double-precision)) → Success: Boolean), the description and category, a table of inputs and outputs with type, default value and “(by ref)”, the local variables in a collapsible list, and the graph image.

Types, parent classes, interfaces, components and references link to the page of the asset they name when it is part of the export. Engine and C++ types are plain text.

Graph viewer

Click a graph image to open it in the viewer. Scroll to zoom, drag to pan, use Fit and 100% to reset the view, and close it with the close button or Escape. See Graph images for how the images are made.

Private members and audit findings

With Private members off, private variables and functions are left out of the pages and the search. With Audit findings off, the Audit link, audit.html, the Audit sections and the Issues and health numbers on the overview are left out.

Markdown

The Markdown files are written to the markdown folder:

• index.md: the title, the scope, the number of assets, the coverage, with audit findings included the health score, error and warning counts and a link to audit.md, and the average complexity with a link to migration.md. Then a table of all assets with Name, Kind, Folder, Nodes, Complexity, C++ and Documented.
• audit.md: a table of all findings with Severity, Asset, Rule and Finding. Only with audit findings included.
• migration.md: the C++ migration report as a table with Asset, C++ score, Verdict, Complexity and Why.
• One <id>.md file per asset with the same content as the HTML page: a facts table, then the C++ migration, audit, entries, components, widgets, variables or fields, functions, events, event dispatchers, macros, graphs, notes and references.

Links between the files are relative (Game.Blueprints.BP_Door.md). Graph images are referenced as ../images/<id>/<graph>.png, so keep the markdown and images folders next to each other when you copy the files, for example into a wiki repository.

JSON report

scribe.json contains the whole scan: every asset with everything Blueprint Scribe found, and every finding. It is UTF-8 without a byte order mark.

Note

The JSON report always contains all members and all findings. The Private members option only affects the HTML, Markdown and AI-ready text output, and Audit findings only the HTML and Markdown output.

Example

An asset from an export (an Animation Blueprint), followed by one finding:

{
  "generator": "ta Blueprint Scribe 1.1.0",
  "title": "Scribe Test",
  "generated": "2026-10-08T21:53:15.369Z",
  "healthScore": 69,
  "docCoverage": 0.75609755516052246,
  "complexityScore": 14,
  "scope": [
    "/Game/ScribeTest"
  ],
  "assets": [
    {
      "name": "ABP_Cube",
      "package": "/Game/ScribeTest/Animation/ABP_Cube",
      "objectPath": "/Game/ScribeTest/Animation/ABP_Cube.ABP_Cube",
      "assetClass": "AnimBlueprint",
      "kind": "Animation",
      "id": "Game.ScribeTest.Animation.ABP_Cube",
      "description": "Animation Blueprint of the engine's skeletal cube.",
      "category": "",
      "displayName": "",
      "namespace": "",
      "parentChain": [
        { "text": "AnimInstance", "target": "", "note": "" }
      ],
      "interfaces": [],
      "components": [],
      "variables": [],
      "functions": [],
      "events": [],
      "dispatchers": [],
      "macros": [],
      "graphs": [
        {
          "name": "AnimGraph",
          "kind": "Anim Graph",
          "path": "AnimGraph",
          "nodeCount": 1,
          "logic": "Nodes: Output Pose",
          "image": "images/Game.ScribeTest.Animation.ABP_Cube/AnimGraph.png",
          "imageSize": { "x": 275, "y": 260 }
        }
      ],
      "notes": [],
      "widgets": [],
      "widgetAnimations": [],
      "enumEntries": [],
      "targetSkeleton": {
        "text": "SkeletalCube_Skeleton",
        "target": "/Engine/EngineMeshes/SkeletalCube_Skeleton",
        "note": ""
      },
      "uses": [
        { "text": "SkeletalCube_Skeleton", "target": "/Engine/EngineMeshes/SkeletalCube_Skeleton", "note": "" }
      ],
      "usedBy": [],
      "compileStatus": "Up to date",
      "metrics": {
        "nodes": 1,
        "largestGraphNodes": 1,
        "largestGraph": "AnimGraph",
        "hardReferencePackages": 1,
        "hardReferenceBytes": 6505,
        "referencers": 0,
        "documentable": 1,
        "documented": 1,
        "branches": 0,
        "loops": 0,
        "casts": 0,
        "mathNodes": 0,
        "tickNodes": 0,
        "tickLoops": 0,
        "complexity": 1
      },
      "migration": {
        "score": 0,
        "verdict": "Fine in Blueprint",
        "reasons": [],
        "upgradeNotes": []
      },
      "docCoverage": 1
    }
  ],
  "issues": [
    {
      "asset": "/Game/ScribeTest/Blueprints/BP_Door.BP_Door",
      "assetName": "BP_Door",
      "rule": "Tick",
      "ruleName": "Event Tick",
      "severity": "Warning",
      "message": "Event Tick runs 24 nodes every frame. Timers, events or a lower tick interval are often cheaper.",
      "graph": "EventGraph",
      "node": "31FC0097-46B1-D39A-D187-ED9EA265DB73"
    }
  ]
}

The example is shortened: the real export has more assets and findings, the Animation Blueprint also lists its empty Event Graph, and the real file has one property per line.

Variable and function

A variable and a function of a Blueprint:

{
  "name": "OpenSpeed",
  "type": { "text": "Float (double-precision)", "target": "", "note": "" },
  "category": "Door",
  "defaultValue": "2.0",
  "description": "",
  "flags": [ "Instance Editable", "Expose on Spawn" ],
  "bInstanceEditable": true,
  "bPrivate": false,
  "bUsed": true
}

{
  "name": "Open",
  "description": "Opens the door at the given speed. Returns false when it is locked.",
  "category": "",
  "access": "Public",
  "flags": [],
  "inputs": [
    { "name": "Speed", "type": { "text": "Float (double-precision)", "target": "", "note": "" }, "defaultValue": "", "bByRef": false }
  ],
  "outputs": [
    { "name": "Success", "type": { "text": "Boolean", "target": "", "note": "" }, "defaultValue": "true", "bByRef": false }
  ],
  "localVariables": [],
  "graphName": "Open",
  "ownerGraph": "",
  "bOverride": false,
  "bPrivate": false,
  "nodeGuid": "2837C84A4FB37403CFBDFFA34CEF5926"
}

Top level

Field	Meaning
generator	ta Blueprint Scribe and the plugin version
title	Title of the export (the project name followed by “Blueprints” when no title is set)
generated	Time of the export in UTC (ISO 8601)
healthScore	Health score from 0 to 100. See Health and coverage.
docCoverage	Documentation coverage of all assets, from 0 to 1
complexityScore	Average complexity of the Blueprints with logic, from 0 to 100
scope	The scanned folders, or the names of the scanned assets
assets	One object per asset, sorted by package path
issues	Every finding of the scan

Assets

Boolean fields keep Unreal’s b prefix, for example bPrivate.

Field	Meaning
name	Asset name
package	Package path, for example /Game/Blueprints/BP_Door
objectPath	Object path, for example /Game/Blueprints/BP_Door.BP_Door
assetClass	Asset class, for example Blueprint, WidgetBlueprint, AnimBlueprint, UserDefinedStruct or UserDefinedEnum
kind	Actor, Component, Object, Widget, Animation, Interface, Function Library, Macro Library, Structure or Enum
id	File name (without extension) of the asset’s pages
description	Blueprint description, structure tooltip or enum description
category, displayName, namespace	Blueprint category, display name and namespace from Class Settings
parentChain	Links to the parent classes, from the direct parent up to (without) Object
interfaces	Links to the implemented interfaces
components	Components of an Actor Blueprint
variables	Member variables, or the fields of a structure
functions, events, dispatchers, macros	Function objects
graphs	Graph objects
notes	Comment boxes and node comments
widgets	Widget hierarchy of a Widget Blueprint
widgetAnimations	Names of the widget animations
enumEntries	Entries of an enum
targetSkeleton	Link to the skeleton of an Animation Blueprint; empty fields otherwise
uses	Links to the packages this asset references: hard references first, then soft ones with note soft. C++ (/Script) packages are left out.
usedBy	Links to the packages that reference this asset. Actors placed in World Partition levels are shown as their level, with note placed in level.
compileStatus	Up to date, Warnings, Error, Dirty or Unknown; empty for structures and enums
metrics	Size, logic and documentation numbers, see below
migration	The C++ migration result: score (0 to 100), verdict (Strong C++ candidate, Consider C++, Fine in Blueprint or No logic), reasons and upgradeNotes
docCoverage	documented ÷ documentable of this asset, from 0 to 1 (1 when there is nothing to document)

Nested objects

Object	Fields
Link	text (the name shown), target (package of the asset it points to; empty for C++ classes and types), note (soft, placed in level or empty)
Variable	name, type (link), category, defaultValue, description (the tooltip), flags, bInstanceEditable, bPrivate, bUsed (false when no graph of the Blueprint reads or writes it)
Function	name, description, category, access (Public, Protected or Private), flags, inputs, outputs, localVariables, graphName (graph of a function or macro), ownerGraph (graph that holds an event), bOverride, bPrivate, nodeGuid (32 hexadecimal digits)
Parameter	name, type (link), defaultValue, bByRef
Component	name, class (link), parent (component it is attached to), description, inheritedFrom (class that adds it; empty for the Blueprint’s own components), depth
Graph	name, kind, path (unique within the Blueprint, for example EventGraph/Collapsed Graph), nodeCount (without comments and reroute nodes), logic (the graph as readable text, see Graph logic), image (relative path, empty without an image), imageSize (x, y)
Note	graph, text, node (title of the commented node; empty for comment boxes)
Widget	name, class (link), depth, bIsVariable
Enum entry	name, displayName, description, value
Metrics	nodes (all graphs, without comments and reroute nodes), largestGraph and largestGraphNodes, hardReferencePackages and hardReferenceBytes (see below), referencers (number of usedBy entries), documentable and documented (see Health and coverage), branches, loops, casts and mathNodes (enabled nodes of each type), tickNodes and tickLoops (nodes and loops Event Tick runs every frame), complexity (see Complexity score; 0 for assets without logic)

Graph kinds are Event Graph, Function, Construction Script, Interface, Macro, Collapsed, Anim Graph, State Machine, State, Conduit and Transition.

hardReferencePackages and hardReferenceBytes count every package that is loaded together with the asset through hard references (recursively, without C++ packages) and their size on disk.

Issues

Field	Meaning
asset	Object path of the asset
assetName	Asset name
rule	Rule id, for example Graph.Large. See Audit rules.
ruleName	Rule name, for example Large graph
severity	Error, Warning or Info
message	The finding
graph	Path of the graph the finding is about; empty if none
node	GUID of the node the finding is about (with hyphens); empty if none

AI-ready text

The ai folder holds compact text for AI assistants: one <id>.md per asset, blueprints.md with all assets and blueprints.jsonl with one JSON line per asset. Every graph is written as readable logic. With only this format selected, ai/blueprints.md is the start file of the export. See AI export.

The Private members option applies to the AI-ready text as well. Audit findings are always included in it.

Credit line

The footer of every HTML page shows “Generated” and the export time. With Credit line on (the default), it continues with “with ta Blueprint Scribe 1.1.0 by tridiART” (with the installed version number), where the plugin name links to tridiart.com. The last line of every Markdown asset file reads “Generated” and the export time, followed by “with ta Blueprint Scribe by tridiART” when the credit line is on.

Turn the credit line off on the Export page, with Show Credit in the settings, or with show_credit in Python. The pages then only show the export time. The generator meta tag of the HTML pages and the generator field of scribe.json are always written.