Karpathy LLM Wiki: The Pattern Behind Agent Memory (Complete Guide)
Andrej Karpathy's LLM Wiki gist (5K+ stars) replaces RAG re-retrieval with a persistent, agent-maintained markdown wiki. Three layers, ingest/query/lint ops, vs OKF and RAG—plus the ecosystem it spawned.
Andrej Karpathy's LLM Wiki gist is one of the most influential agent-memory documents of 2026—5,000+ stars and 5,000+ forks on a single markdown file that fits on one screen. It is not a product. It is an idea file meant to be copy-pasted into Claude Code, Codex, OpenCode, or Pi so your agent builds the specifics with you.
The core reframe:
Most LLM + document workflows are RAG: upload files, retrieve chunks, generate an answer. The LLM rediscovers knowledge from scratch every time. Nothing accumulates.
LLM Wiki is different: the LLM incrementally builds and maintains a persistent wiki—structured, interlinked markdown between you and raw sources. Knowledge is compiled once and kept current, not re-derived on every query.
Google formalized a vendor-neutral version as Open Knowledge Format (OKF) days after the gist went viral. OpenKnowledge from Inkeep (June 2026) ships a WYSIWYG editor and agent wiring (ok init) as a product implementation of the same pattern. This guide covers Karpathy's original pattern, the three layers, the ingest/query/lint loop, when wiki beats RAG, and the ecosystem the gist spawned.
Wiki wins below ~50K–100K tokens; RAG for millions+
The Problem With RAG-Only Workflows
NotebookLM, ChatGPT file uploads, and most enterprise RAG share a limitation Karpathy names explicitly:
Ask a subtle question requiring synthesis across five documents, and the LLM must find and piece together fragments every time. Cross-references are not pre-built. Contradictions are not pre-flagged. Synthesis does not compound.
LLM Wiki treats maintenance as the LLM's job:
When you add a source, the agent integrates it—updates entity pages, revises summaries, notes contradictions
Cross-references already exist when you query
The wiki gets richer with every source and every good answer you file back
Karpathy's workflow metaphor: Obsidian is the IDE; the LLM is the programmer; the wiki is the codebase. You browse graph view and links on one side; the agent edits on the other.
Three Layers
1. Raw sources (immutable)
Your curated collection: articles, papers, images, data files, meeting transcripts, code exports. The LLM reads them but never modifies them. This is source of truth.
Tip from the gist:Obsidian Web Clipper converts web articles to markdown. Set a fixed attachment folder and bind "Download attachments for current file" so the LLM can reference images locally.
2. The wiki (LLM-owned)
A directory of LLM-generated markdown:
Page type
Purpose
Source summaries
One page per ingested document
Entity pages
People, companies, concepts
Topic summaries
Evolving synthesis
Comparisons / analyses
Filed from query operations
overview.md
High-level map of the domain
You read the wiki; the LLM writes and maintains it—including cross-links, consistency, and updates when new sources arrive.
3. The schema (CLAUDE.md / AGENTS.md)
The configuration file that makes the LLM a disciplined wiki maintainer rather than a generic chatbot:
Directory structure conventions
Page formats and naming
Ingest workflow (steps after new source)
Query workflow (search, cite, file good answers back)
Lint workflow (health checks)
You and the LLM co-evolve this file as you learn what works for your domain. See What is CLAUDE.md? for how this fits Claude Code specifically.
Three Operations
Ingest
Drop a source in raw/, tell the LLM to process it.
Typical flow:
LLM reads the source
Discusses key takeaways with you (optional but Karpathy prefers staying involved)
Writes a summary page in the wiki
Updates index.md
Updates relevant entity and concept pages (often 10–15 files per source)
Appends entry to log.md
Karpathy prefers one source at a time with human review. Batch ingest with less supervision is possible—document your choice in the schema.
Query
Ask questions against the wiki, not raw files:
LLM reads index.md to locate relevant pages
Drills into pages
Synthesizes answer with citations
Answers can take many forms: markdown, comparison tables, Marp slide decks, matplotlib charts. Critical insight: good answers should be filed back into the wiki as new pages—explorations compound like ingested sources.
Lint
Periodically health-check the wiki:
Check
Action
Contradictions between pages
Flag or reconcile (domain-dependent)
Stale claims superseded by newer sources
Update or mark superseded
Orphan pages (no inbound links)
Link or merge
Concepts mentioned but no dedicated page
Create stub pages
Missing cross-references
Add links
Data gaps
Suggest web search or new sources
The LLM suggests new questions to investigate—lint is proactive, not just cleanup.
index.md vs log.md
File
Orientation
Purpose
index.md
Content
Catalog of all pages—link, one-line summary, optional metadata (date, source count). Updated on every ingest. Query entry point.
log.md
Chronological
Append-only timeline of ingests, queries, lint passes
Chapter-by-chapter filing → personal Tolkien Gateway
Business/team
Slack, meetings, docs → LLM-maintained internal wiki
Competitive analysis
Due diligence, market maps
Trip planning, courses, hobbies
Any accumulating knowledge
Karpathy links the idea to Vannevar Bush's Memex (1945)—private, curated knowledge with associative trails. Bush couldn't solve maintenance; the LLM handles that.
LLM Wiki / full context — 100% retrieval reliability, no vector DB, global reasoning
Millions of tokens+
RAG — won't fit in context
In between / production
Hybrid — stable core in wiki, dynamic mass in RAG
Important nuance: index.md is not RAG. It does not vector-match or chunk—it lets the agent open fewer whole files. Even reading the entire library often fits modern 200K–1M token windows; the index is an optimization.
They stack: CLAUDE.md can point agents at an OKF bundle or LLM wiki directory and define ingest/query/lint rules.
Why This Works (Karpathy's Argument)
The tedious part of maintaining a knowledge base is not the reading or the thinking — it's the bookkeeping. Updating cross-references, keeping summaries current, noting when new data contradicts old claims.
Humans abandon wikis because maintenance cost grows faster than value. LLMs don't get bored, don't forget cross-references, and can touch 15 files in one pass.
Your job: curate sources, direct analysis, ask good questions, think about meaning.
The LLM's job: everything else.
Summary
Karpathy's LLM Wiki is the clearest statement yet that agent memory should compound—not re-retrieve. Three layers (raw → wiki → schema), three operations (ingest → query → lint), two navigation files (index.md, log.md), and a git repo of markdown as the artifact.
Copy the gist into your agent. Let it build the rest with you. For organizational interoperability, layer Google's OKF on top when you need cross-team bundles.
The wiki is just markdown in git. Version history, branching, and collaboration come free.
