Appendix Guide
What this is: the appendix area holds content that does not belong to the pyramid argument line but still must have a place — training bridges, real-world cases, course notes, and historical methodology archives. They answer "what else is worth reading", not "what to learn next".
Boundary statement: appendices do not participate in the pyramid argument order. The mainline runs from the tech map through layers 0→5; appendix pages can be read at any time but are never cited as prerequisites by mainline chapters. Bridge pages (status: bridge) outsource deep theory to Learn LLM; case pages (status: case) supply real-world evidence; legacy pages preserve content that the mainline has absorbed or superseded.
Sections
| Section | One-liner | Status | Language side |
|---|---|---|---|
| Case studies | SLS log assistant, golden dataset, testing cases and more real-world write-ups | case | some pages exist on one side only — see markers |
| Course notes | Agent Course and other external course/article notes | legacy | some pages exist on one side only — see markers |
| Methodology archive | AI-assisted programming, workflow automation, n8n, Midscene archives | legacy | some pages exist on one side only — see markers |
| AI coding tool cases | Cursor architecture and rules, Copilot system prompt excerpts | case | body on the ZH side |
Reading advice
- Solving a current problem: go back to the tech map and follow the symptom-driven decision tree; do not start from appendices.
- Want model-side theory (training, alignment, multimodal internals): go straight to Learn LLM; this repo keeps only engineering-decision bridges.
- Want to see how others did it: enter case studies and filter by your layer.
- Tracing the provenance of this repo's content: course notes and the methodology archive keep links to original sources.