Specifications, tool development, and thoughts on the future of AI.
Starting with how society specializes and collaborates today, explore how AI might develop and assess which projects fit the changes underway and can be built now.
Support project growth with clear, executable, evolving specifications, from code architecture to AI collaboration.
Scope, condition, response: what does a check rule need to explain?
Build each iteration on what already works.
ScriptSlateBridge / PythonSlate
Build UE editor interfaces in Python with shared widgets, menus, and editor capabilities, giving different tools a common foundation.
My focus: making recurring UI patterns and interactions reusable, with clear, stable boundaries for building new tools with AI.
SceneCheckToolkit
Organize checks around scenes and components in the editor, connecting usage context, issue location, and actionable feedback.
My focus: preserving object relationships as checks move from assets into scenes, so results can lead back to a specific usage location.
ResourceToolkit
Combine scope, conditions, object transformations, and actions into reusable rules that bring production standards into everyday workflows.
Read about rule designAssetCheckToolkit
Organize checks around assets, import and display results, and help users locate issues, understand findings, and take the next step.
My focus: how people review and use results, and how tools can share common property and UI capabilities.
Yingying
An Android AI chat app I am developing. Users ask a question, and the server selects the reply model and reasoning effort based on the conversation, keeping the exchange in the same chat.
Sign in with an account. Model calls are billed by actual usage; the app explains the rules and prices.
The core library provides common capabilities; each tool owns its domain logic and workflow.
Editor tools are not available for public download yet. Yingying's Android package is available above.
Start with one rule: evaluate, act, and check again.
This example uses fictional texture sizes: define the requirement, check the actual value, adjust it, and verify again. Real tools also need to handle data sources, limits on changes, and location information.
| Demo asset | Max. dimension | Result |
|---|---|---|
| T_DemoStone_BC | 4,096 | Exceeds limit |
| T_DemoTile_N | 2,048 | Pass |
| T_DemoIcon | 512 | Pass |
Fictional data illustrating rule logic only. This page does not connect to UE or change real assets.
My work centers on large-world game development in UE5 and editor tools. I continue to develop a shared Python editor library, standards validation, and asset checks, while refining tools for scene checks.
I explore how AI can shorten the path from a requirement to a first implementation, and how architecture, shared capabilities, and verification support long-term maintenance. These notes cover concrete problems, implementation trade-offs, and improvements informed by real use.
Get in touch · Email29797471@qq.com