Suggestion: add WFGY 16 Problem Map as a practical debugging map for RAG / long-horizon prompts

[onestardao]

Hi, thanks a lot for putting this collection together – it is very helpful to see in-context learning and prompt engineering resources in one place.

I wanted to share one practical resource that might complement the more theoretical papers you list, especially for people who are building real RAG or long-horizon workflows.

What it is

• Name: WFGY 16 Problem Map
• Link: https://github.com/onestardao/WFGY/tree/main/ProblemMap/README.md
• Type: vendor-neutral debugging map for LLM pipelines (focus on RAG, vector stores, long context prompts)
• License: MIT

The map is written as a set of 16 failure modes, each with:

• a short description of the failure pattern
• observable symptoms (what users actually see)
• a minimal structural fix (not just “tune the prompt”)

Examples include ingestion drift, chunking/contract mismatch, vector store metric issues, retrieval ordering bugs, and evaluator mismatch for long chains.

If you think it fits, maybe it could live in a section like “Practical debugging resources” or “RAG / system-level techniques”.

If this feels out of scope, no worries at all – just wanted to surface it in case it is useful to your readers. Thanks again for curating this repo.

[onestardao]

Hi, just a brief follow-up on this suggestion.

Since opening this issue, the WFGY 2.0 “RAG 16 Problem Map” has been adopted or referenced by 20+ repositories across the RAG / LLM ecosystem, including:

• RAGFlow
• LlamaIndex
• ToolUniverse (Harvard MIMS Lab)
• Rankify (University of Innsbruck)
• Multimodal RAG Survey (QCRI LLM Lab)

It has also been included in multiple curated “awesome” lists and LLM resource collections as a structured RAG failure-mode checklist.

In these integrations, WFGY is not positioned as a competing framework. It functions as a vendor-neutral diagnostic layer — a structured failure vocabulary for debugging and evaluating RAG / LLM pipelines.

Given that this repository curates tools and references around LLM systems, I wanted to check again whether adding WFGY as a debugging / evaluation resource would align with your scope.

If so, I’m happy to submit a minimal MVP PR — just a concise one-line entry formatted exactly to your style.

If it’s outside the intended focus, no worries at all. Just wanted to confirm before proceeding.

Thanks again for maintaining this list.