Proposal: add a WFGY 16-problem RAG failure checklist for Redis Stack / RedisVL based RAG #520

Hi Redis team,

Redis Stack and RedisVL are frequently used as the vector and cache layer for RAG applications. When answers are wrong or unstable, users often do not know whether the problem is data, configuration, or higher level pipeline design.

I maintain WFGY RAG 16 Problem Map, an open-source failure map for RAG and LLM pipelines.

Repo (MIT):
https://github.com/onestardao/WFGY

Main reference page:
https://github.com/onestardao/WFGY/tree/main/ProblemMap/README.md

The project offers:

• A list of 16 common RAG failure modes
• A one-page Global Debug Card
• A triage prompt that uses a failing run to classify the failure and propose structural fixes

The same map is already integrated or cited in:

• RAGFlow and LlamaIndex RAG troubleshooting docs
• ToolUniverse (Harvard MIMS Lab) as a triage tool
• Rankify (University of Innsbruck)
• Multimodal RAG Survey (QCRI LLM Lab)
• Curated lists such as Awesome LLM Apps and Awesome Data Science – academic

Proposal:

Add a short RAG failure checklist for Redis Stack / RedisVL based RAG. For example:

1. A docs page that:

   • Explains typical failure scenarios when Redis is used for vector search and caching.
   • Maps these scenarios to WFGY’s 16 failure types.
   • Links to the Global Debug Card so users can run the triage prompt on their own failing runs.

2. An example where:

   • A Redis based RAG pipeline is built and intentionally misconfigured.
   • Failing calls are captured with queries, retrieved vectors, and logs.
   • The WFGY triage prompt is used to label each failure and guide fixes.

This would give Redis users a structured diagnostic tool that is compatible with many different RAG frameworks while keeping Redis itself focused on performance and reliability.

If you like this idea I can help draft the guide or example in your preferred format.