Skip to content

Index and Log

Summary

Two special files that help the LLM (and humans) navigate the wiki as it grows: index.md is a content-oriented catalog of all pages, while log.md is a chronological record of all operations performed.

Index.md — Content Catalog

Organized by category (concepts, entities, sources, insights), each entry includes: - Link to the page - One-line summary of what it covers - Metadata: source count, last updated date, tags

Purpose: When answering a query, the LLM reads the index first to find relevant pages, then drills into them. This works surprisingly well at moderate scale (~100 sources, ~hundreds of pages) and avoids the need for embedding-based RAG infrastructure.

Karpathy noted: "I thought I had to reach for fancy RAG, but the LLM has been pretty good about automaintaining index files."

Log.md — Chronological Record

Append-only, newest entries at the bottom. Each entry starts with a consistent prefix for parsability:

```

[YYYY-MM-DD] ingest | Source Title

  • Type: article | paper | data | podcast
  • Pages touched: [[Page A]], [[Page B]], ...
  • Key insight: one-line summary ```

Purpose: Gives a timeline of the wiki's evolution. Helps the LLM understand what's been done recently. Parseable with simple unix tools — grep "^## \[" log.md | tail -5 gives the last 5 entries.

Two Different Purposes

index.md log.md
Orientation Content-oriented Chronological
Question it answers "What pages exist on topic X?" "What happened recently?"
Structure Categorized table Timestamped entries
Update pattern Regenerated on every ingest Append-only, never modified

See Also