WFGY (LLM debugging and long horizon TXT stress test) as a candidate entry

[onestardao]

Hi, thanks for maintaining this list. It has been very helpful as a map of the LLM ecosystem.

I wanted to ask whether a project called WFGY might be in scope for Awesome LLM, probably under the sections related to evaluation, debugging, or reasoning frameworks.

Very short overview of the three public stages:

• WFGY 1.0 is a PDF that treats large language models as self-repairing systems. It defines a small loop of modules that watch the conversation and correct drift at the text layer, without touching model weights. The PDF includes experimental results on standard benchmarks and a complete reasoning workflow.

• WFGY 2.0 builds on that by introducing DT-related formulations and a curated list of 16 engineering-level problem cases. The focus is on practical failure modes such as RAG pipelines, vector databases, and deployment ordering issues. Its role is diagnostic and corrective for real systems, not just conceptual discussion.

• WFGY 3.0 · Singularity Demo is a TXT pack that compresses the accumulated structure into a form that models can directly read. After uploading the file, the model shows a small menu. Following the on-screen flow and answering go at the right time lets the model replay the same evaluation trace, which different systems can reproduce and compare.

All of this lives in a single open source repo on GitHub (MIT license, 1k+ stars). The intention is not to be another wrapper, but to be a debugging and stress-testing layer that can sit on top of any base model.

If you feel WFGY fits the scope of Awesome LLM, I am happy to prepare a small PR that adds one concise entry in the most appropriate section. If you prefer to keep the list focused on other categories, that answer is also totally fine. I mainly wanted to ask before sending a PR.

Thanks again for curating this project.