Hyper-Extract vs llm_wiki
Knowledge-extraction CLI: LLMs turn documents into structured graphs, hypergraphs and spatio-temporal knowledge — with an MCP server for agents and Obsidian vault export. — versus — Desktop app implementing Karpathy's LLM Wiki pattern: an LLM ingests your documents into a persistent, interlinked wiki with a knowledge graph, hybrid search, deep research and an MCP server.
Both compile documents into linked knowledge: hyper-extract is a CLI emitting graphs and Obsidian vaults for agents; LLM Wiki is a GUI app that keeps its wiki current as sources change.
| Hyper-Extract | llm_wiki | |
|---|---|---|
| Stars | 4.0k | 20k |
| Forks | 457 | 2.3k |
| Language | Python | TypeScript |
| License | Apache-2.0 | NOASSERTION |
| Last activity | 8 days ago | yesterday |
| Topics | knowledge-graphs, rag | rag, knowledge-graphs |
| Curated connections | 10 | 4 |
Hyper-Extract — the curator's take
The pitch beyond ordinary KG extraction is the hypergraph: relations that connect MORE than two entities survive instead of being flattened into pairwise triples. One command per document, query the abstracts over MCP from Claude Desktop or your IDE, export to Obsidian wikilinks. NOT a graph database (it extracts, storage stays simple) and no standard license resolution at review time — verify before building on it; extraction quality tracks the LLM you plug in.
llm_wiki — the curator's take
Pick LLM Wiki if you like Karpathy's 'compile knowledge once instead of re-retrieving it every query' idea but don't want to wire it up yourself: drop in PDFs, Office docs, EPUBs or web clips and an LLM writes and maintains source-traced wiki pages, with Louvain clusters, gap-finding and deep research on top, plus an MCP server and a skill so Claude Code or Codex can query it. It's a personal, single-user desktop app — not a team knowledge service (see arkon) — and every ingest spends LLM tokens on a two-step analyse-then-write pass, so big corpora cost real money. Licence is GPL-3.0 (the GitHub API reports NOASSERTION): fine to use, copyleft if you fork and redistribute.