Bug report 03
Embedded Documentation and Naming Contract Failure
The problem
Large uploaded apps can be self-documenting. This project embeds build metadata, naming conventions, delivery workflow, and human requirements inside the source. A model that patches code without honoring those embedded rules can generate invalid artifacts even when the code change itself is useful.
Concrete failure class
| Expected | Failure |
|---|---|
| Follow BUILD_VERSION and BUILD_FILE naming. | Invented a package name outside the app convention. |
| Use ZIP-only delivery when documented. | Exposed raw HTML as a standalone chat artifact. |
| Read the ledger before modifying behavior. | Treated the file like ordinary small HTML. |
Requested product fix
When a source file contains explicit build metadata, artifact naming, or delivery contract comments, ChatGPT should surface and follow those rules before generating files.
2000-character report
ChatGPT must respect documentation embedded inside uploaded source files before changing conventions. This project is a single-file app whose source contains build metadata, naming rules, delivery workflow, and human requirements. The assistant skimmed enough to patch code but invented an artifact name and exposed raw HTML, violating documented workflow. For large self-documenting apps, ChatGPT should read and follow embedded build constants and requirements ledgers before generating artifacts. The model should preserve file naming, use ZIP-only delivery when documented, and avoid treating huge HTML apps like ordinary small snippets.