Knowledge Management

How to collect, organize, and compound knowledge over time. The tools

and systems for building a personal knowledge base that grows more

valuable the larger it gets.

Key Entries

• **Johnny Decimal System** -- numerical categorization for files and

information. Strict hierarchy, easy to remember, scales to thousands.

• **Gitingest** -- convert GitHub repos into LLM-digestible text. Turning

codebases into knowledge.

• **Crawl4AI** -- web crawler optimized for LLM ingestion. The front door

of the knowledge pipeline.

• **Karpathy on Auto Research** -- automated research workflows using LLMs.

The vision: agents that read papers and build summaries.

• **Org-mode** (implicit) -- the substrate this entire knowledge base is

built on. Plain text, hierarchical, tag-based, infinitely extensible.

The Karpathy Pattern

raw/ -> LLM compiles -> wiki/ -> query -> enhance -> repeat

This knowledge base follows exactly this pattern:

1. Raw data ingested from org links and HTML archives

2. LLM compiles into categorized wiki with concept articles

3. Wiki is queryable and browsable

4. Queries and explorations feed back into the wiki

5. The knowledge base compounds

The key insight: the LLM maintains the wiki, not you. You curate

inputs and ask questions. The LLM does the filing, linking, and

synthesis.

Connections

• Links to AI Agent Frameworks -- agents

as knowledge workers

• Links to Privacy-First Tools -- your

knowledge base should be private by default

• Links to The Solo Builder -- knowledge

management as a competitive advantage for solo operators