📌 This is just an index repo.
Go to each problem’s folder for detailed modules, demos, and engineering walkthroughs.
👉 Main Problem Map

• Problem Map 1.0 — Core 16 Failures
• Problem Map 2.0 — RAG Architecture & Recovery
• Grandma Clinic — All Bugs Made Simple (1–16)
• Semantic Clinic — Triage Hub
• Diagnose by Symptom — Quick Sheet
• 🧩 MVP Demos — Minimal Runnable Examples
• 🌍 Global Fix Map — Systemic AI Failures

This is the central index of AI failure modes we've tracked, diagnosed, and solved using WFGY — a semantic reasoning engine for LLMs.

We don't just identify bugs. We redefine them as solvable cognitive flaws — then ship modular logic patches you can drop into your system.

If your LLM has ever hallucinated, frozen, collapsed, or forgotten... chances are we've dissected it here.

Each link opens the failure’s fix page:

1. Hallucination & Chunk Drift — retrieval returns wrong/irrelevant content
2. Interpretation Collapse — chunk is right, logic is wrong
3. Long Reasoning Chains — multi-step tasks silently drift
4. Bluffing / Overconfidence — confident but unfounded answers
5. Semantic ≠ Embedding — cosine match ≠ true meaning
6. Logic Collapse & Recovery — dead-end logic; needs controlled reset
7. Memory Breaks Across Sessions — lost threads, no continuity
8. Debugging is a Black Box — no visibility into failure path
9. Entropy Collapse — attention melts, incoherent output
10. Creative Freeze — flat, literal outputs
11. Symbolic Collapse — abstract/logical prompts break
12. Philosophical Recursion — self-reference/paradox loops
13. Multi-Agent Chaos — agents overwrite/misalign logic (overview)
14. Bootstrap Ordering — services fire before deps ready
15. Deployment Deadlock — circular waits (index⇆retriever, DB⇆migrator)
16. Pre-Deploy Collapse — first call after deploy crashes

Multi-Agent deep dives

• Role Drift
• Cross-Agent Memory Overwrite

• Query Parsing Split (HyDE/BM25)
• Symbolic Constraint Unlock (SCU)
• Hallucination Re-entry
• Memory Desync
• Vectorstore Fragmentation
• Bootstrap Deadlock (RAG boot fence)

All patterns: Patterns Index

• Retrieval Playbook
• Rerankers
• Data Contracts
• Multilingual Guide
• Privacy & Governance

Examples: Examples Hub
Eval: Evaluation Suite
Ops Runbook: Ops & Monitoring
Beginner Guide: Start Here
Getting Started: RAG Quickstart

• Map-A · RAG Problem Table
• Map-B · Multi-Agent Chaos Map
• Map-C · Symbolic & Recursive Map
• Map-D · Logic Recovery Map
• Map-E · Long-Context Stress Map
• Map-F · Safety Boundary Map
• Map-G · Infra Boot Map

• Demo Readme & Entrypoints
• Browse All Demos

👑 Early Stargazers: Hall of Fame — Engineers, hackers, and open-source builders who supported this project from day one.
View the Hall of Fame