unreal-blueprint-analyzer

Unreal Blueprint Analyzer

Read-only analysis of Blueprint binary assets in any Unreal Engine project. Treat .uasset, .umap, and similar Unreal assets as binary source files; never patch them in place.

Core rules

• Do not edit .uasset, .umap, .navmesh, or other Unreal binary assets directly.
• Separate confirmed fact from inference. Label evidence sources as source, docs, references, binary strings, or editor exports.
• Prefer the most reliable source available: official editor exports, Commandlets, user-provided screenshots, or JSON.
• If the user asks to modify a Blueprint, give safe editor steps or a scriptable editor/export pipeline, not a binary patch.

Parsing method

Prefer structured parsers over raw string grepping:

• CUE4Parse (the .NET library behind FModel) for .uasset, .umap, and .pak package structure.
• UAssetAPI for Blueprint Kismet bytecode and expression graphs.
• FModel as a ready-made viewer and exporter.

Fall back to python <skill>/scripts/extract_uasset_strings.py <asset> --json (or strings / PowerShell byte reads) when no structured parser is available.

Workflow

1. Identify the asset path and type: Actor Blueprint, UI Widget, Behavior Tree Task, Data Asset, Map, or unknown.
2. Find adjacent source, config, and docs: match the asset name or path to nearby C++, Blueprint nativization output, Python/Editor Utility, Verse, Lua, C#, plugin scripts, config files, data tables, or project scripting layers; search asset names, generated class names, parent classes, ClassPaths, component names, Widget names, event/function names, Delegates, RPC names, Gameplay Tags, Blackboard keys, Input Actions, Data Table rows, and config references; read AGENTS.md, CONTEXT.md, README.md, design docs, plugin docs, or source comments if present.
3. Extract clues without modifying the asset: prefer the built-in script python <skill>/scripts/extract_uasset_strings.py <asset> --json; otherwise use strings, PowerShell byte reads, or editor exports.
4. Classify the extracted clues: asset paths, ClassPaths, and parent classes; component names, Widget names, and exposed variables; function, event, and RPC names; Blackboard keys, Data Table names, tags, sockets, animation names, and referenced assets.
5. Cross-validate clues against source, config, docs, and editor exports. Strings that appear only in the binary are weak evidence; clues that also appear in source, config, docs, or exports are stronger.
6. Write a short report. When static analysis cannot prove execution flow, give a confidence level and concrete editor re-check points.

Sunset note

Blueprint is still a first-class authoring system in shipping UE5, but Epic has announced a gradual migration to Verse (UE6 early access expected Q4 2027, full by mid-2029). Treat Blueprint assets as long-lived but not evergreen; do not present this workflow as permanent.

Output format

Unless the user specifies otherwise, use this structure:

• Asset: path, inferred type, and related source/config/export paths.
• Purpose: what the Blueprint most likely does, with a confidence level.
• Evidence: grouped into confirmed source/doc evidence and binary-string inference.
• Runtime contracts: component names, Widget names, fields, Blackboard keys, RPCs, asset paths, and save/sync fields the runtime depends on.
• Risks / Unknowns: what static binary analysis cannot prove.
• Editor follow-up: specific items to re-check in the Blueprint editor, or safe manual changes.

Local command examples

python .\scripts\extract_uasset_strings.py <path-to-asset>.uasset --json
rg "BP_Door|Door_C|OpenDoor|Interact|BeginOverlap" Source Content Config Plugins docs

When a project uses a custom Unreal tech stack, follow that stack for adjacent code — do not presume a scripting language. For example: C++ classes in Source/, Editor Utility Python, Verse in UEFN projects, Lua in Oasis/PUBG Mobile UGC projects, C# in UnrealCLR-style integrations, plugin-defined Blueprint Libraries, or data-driven config.