# mindbase [Health: Active] **Category:** ๐Ÿ’ป Developer Tools **Repository:** https://github.com/frankchu91/mindbase **GitHub Stars:** 95 **Views:** 0 **Installs:** 0 **Upvotes:** 0 **Directory Page:** https://allmcps.com/mcp/mindbase ## Description LLM-maintained personal wiki: cross-references sources, flags contradictions. Local markdown. ## Claude Desktop Quick Installation Install path detected from listing signals. Uses `npx` (confidence: high): ```json "mcpServers": { "mindbase": { "command": "npx", "args": ["-y","mindbase-app"] } } ``` ## Documentation & README # MindBase โ€” Karpathy's LLM Wiki, as a product [![npm](https://img.shields.io/npm/v/mindbase-mcp?label=mindbase-mcp)](https://www.npmjs.com/package/mindbase-mcp) [![npm downloads](https://img.shields.io/npm/dw/mindbase-mcp)](https://www.npmjs.com/package/mindbase-mcp) [![CI](https://github.com/frankchu91/mindbase-llm-wiki/actions/workflows/ci.yml/badge.svg)](https://github.com/frankchu91/mindbase-llm-wiki/actions) [![License: MIT](https://img.shields.io/badge/license-MIT-green)](LICENSE) [![Website](https://img.shields.io/badge/website-live%20demos-blue)](https://frankchu91.github.io/mindbase-llm-wiki/) [![Glama score](https://glama.ai/mcp/servers/frankchu91/mindbase/badges/score.svg)](https://glama.ai/mcp/servers/frankchu91/mindbase) > **An open-source implementation of Andrej Karpathy's LLM Wiki idea: an AI that builds and maintains a wiki from your sources.** Not RAG-in-a-vector-DB. A real markdown wiki on your disk, that an LLM gardens for you between conversations. ```bash npx mindbase-app ``` One command: starts the local server, opens the web app, and walks you through picking a **free local model** that fits your RAM. No API key, nothing leaves your machine. (Node 20+)

Demo: /contribute in the MindBase web UI โ€” the AI shows takeaways and a checkbox plan, and writes to the LLM wiki only after approval

MindBase implements Andrej Karpathy's [LLM-Wiki pattern](https://x.com/karpathy/status/1911080091498963196): you feed it sources (papers, articles, thoughts); the LLM reads, cross-references, flags contradictions, and writes structured wiki pages. Later, when you ask a question, the wiki already has the synthesized answer โ€” no vector-search re-derivation at query time. **Status:** Early access, actively developed. What's new: [CHANGELOG](https://github.com/frankchu91/mindbase/blob/HEAD/CHANGELOG.md) ยท [Releases](https://github.com/frankchu91/mindbase-llm-wiki/releases) --- ## Why MindBase You read a lot. Papers, articles, tweets, docs. You want to remember them, connect them, form opinions from them. Today you have two bad options: - **Notion / Obsidian / Roam:** Passive containers. You do all the organizing. AI features are bolted-on generation, not maintenance. - **NotebookLM / Perplexity Pages / ChatGPT search:** RAG-based. Nothing accumulates. Every question re-derives the answer from raw sources. **MindBase is the third option:** the LLM actively maintains a persistent, structured wiki as you feed it sources. Knowledge compounds. Your `context.md` gets sharper every time you contribute. The AI *remembers you across sessions* because your beliefs are written down in markdown files โ€” not stored in a chat history that gets summarized away. Think of it as **a personal Wikipedia that an AI intern writes for you**, kept up to date, cross-referenced, and honest about what it doesn't know. ## How it works (30 seconds) Three physical layers on disk (Karpathy's model):

MindBase: you feed sources, the LLM gardens them into a wiki that compounds โ€” plain markdown on your disk

| Layer | Who owns it | What lives there | |---|---|---| | `sources/` | **You** โ€” append-only, the AI never rewrites it | Quick captures, full notes, PDFs, URLs | | `context.md` + `sources/research/` | **The AI** โ€” every change human-approved | The maintained wiki: synthesis, concept pages, `[[wikilinks]]` | | `state/` ยท `logs/` ยท `artifacts/` | Derived โ€” always rebuildable | Search index, snapshots, lint findings, op history | Three operations run the loop: **ingest** (AI reads a source, discusses takeaways, you approve the wiki updates), **build** (regenerate `context.md` from everything unbuilt), **lint** (the AI audits its own wiki for contradictions, stale claims, and orphans). ## What makes it different - **The AI asks before writing.** Every ingest shows takeaways + a checkbox plan; only what you approve gets written. No black-box edits to your knowledge. - **You can watch the wiki absorb your notes.** Every note carries a status chip โ€” โœจ *Add to wiki* until digested, โœ“ *In wiki* after. - **It audits its own knowledge.** One command re-reads the whole wiki and reports contradictions with the exact conflicting sentences quoted. Notion and NotebookLM structurally cannot do this. - **Free and local by default.** Hardware-detect wizard installs the best Ollama model for your RAM. Cloud keys optional. - **Plain markdown on disk.** Grep it, git it, open it in Obsidian, leave anytime. ## Install **Browser (fastest):** `npx mindbase-app` โ€” shown above. Everything runs locally at `localhost:4321`. **Claude Code (flagship):** the full Karpathy 8-step ingest with sub-agents, slash commands, and per-agent tool boundaries: ``` /plugin marketplace add frankchu91/mindbase-llm-wiki /plugin install mb@mindbase ``` Restart when prompted, then type `/` โ€” you should see `/mb:contribute`, `/mb:build`, `/mb:ask`, `/mb:lint` and 8 more. You get 5 sub-agents with strict tool allowlists (the builder has no file-write tool at all โ€” only an atomic-write MCP call), plus a SessionStart hook that auto-injects your project context.

The core flow in Claude Code: /mb:contribute with an arXiv link โ€” download, archive, discuss takeaways, approve, commit

Cursor Add to `~/.cursor/mcp.json`: ```json { "mcpServers": { "mindbase": { "command": "npx", "args": ["-y", "mindbase-mcp"] } } } ``` Restart Cursor โ€” the tool picker should list `mindbase_contribute` and 48 others. Recommended: add a conventions block to `~/.cursor/rules.md` so the LLM reliably routes "add to mindbase X" to the tools โ€” copy it from the [guide](https://github.com/frankchu91/mindbase/blob/HEAD/docs/guide.md#the-llm-in-cursorwindsurf-ignores-my-add-to-mindbase-request).
Windsurf Cascade settings โ†’ MCP โ†’ add: ```json { "mcpServers": { "mindbase": { "command": "npx", "args": ["-y", "mindbase-mcp"] } } } ``` Same rules-file approach as Cursor works for Cascade.
Cline (VSCode) Cline settings โ†’ **MCP Servers**: ```json { "mcpServers": { "mindbase": { "command": "npx", "args": ["-y", "mindbase-mcp"] } } } ``` Cline auto-detects; every tool call gets a confirmation dialog by default.
Continue.dev `~/.continue/config.json`, under `experimental.modelContextProtocolServers`: ```json { "experimental": { "modelContextProtocolServers": [ { "transport": { "type": "stdio", "command": "npx", "args": ["-y", "mindbase-mcp"] } } ] } } ``` MCP tools appear under `@` in chat.
Any other MCP client (Zed, Aider, Goose, Claude Agent SDKโ€ฆ) Point it at `npx -y mindbase-mcp` as a stdio server. See your client's MCP docs for the config location.
**Next:** create your first project and learn the four daily workflows in the **[Guide โ†’](https://github.com/frankchu91/mindbase/blob/HEAD/docs/guide.md)** ## The web UI Since 0.3 the browser app stands on its own โ€” write notes in a full WYSIWYG editor (`Cmd+N`), quick-capture from anywhere (`Cmd+I`), and run the AI operations with approval cards: `/contribute`, `/build`, `/lint`, `/research`. Live demos on the **[website](https://frankchu91.github.io/mindbase-llm-wiki/)**.

MindBase web UI โ€” category tree on the left, the LLM-maintained context.md in the center, chat starters on the right

**Free local models:** the setup wizard detects your hardware and installs what fits โ€” `llama3.2:3b` (8GB), `qwen3:14b` (24GB+), or Meta's **Muse Glimmer 30B** (32GB+ Apple Silicon, Ollama โ‰ฅ 0.32.7). Measured guidance: qwen3:14b for interactive work (~30s), Glimmer for background lint/build โ€” slower, but its findings quote the exact conflicting sentences. The model switcher on the chat composer flips between them in two clicks.
Feature matrix by editor | Feature | Claude Code | Cursor / Windsurf / Cline / Continue | Web UI | |---|---|---|---| | Slash commands (`/mb:*`) | โœ… | โŒ (use natural language) | โœ… (`/contribute`, `/build`, `/lint`, `/research`) | | MCP tools directly | โœ… | โœ… | โŒ | | Karpathy 8-step ingest with approval | โœ… (sub-agents) | โš ๏ธ Manual via prompt | โœ… (approval cards) | | Contribute / ingest PDF & URL | โœ… | โœ… | โœ… (upload + โœจ Process) | | Ask wiki with cited answers | โœ… | โœ… | โœ… | | Build / health check | โœ… | โœ… | โœ… | | Wiki tree browsing + rich editor | โŒ | โŒ | โœ… | | `-p project-id` routing | โœ… | โš ๏ธ natural language | โœ… (switcher) |
## Where your data lives Everything is plain markdown under `~/mindbase-data/` (override: `MINDBASE_DATA_DIR`): ``` ~/mindbase-data/projects/my-research/ โ”œโ”€โ”€ README.md # Ops manual โ€” you edit, LLM reads โ”œโ”€โ”€ context.md # Synthesized truth โ€” LLM writes, you approve โ”œโ”€โ”€ index.yaml # Auto-generated catalog โ”œโ”€โ”€ sources/ โ”‚ โ”œโ”€โ”€ contributors// # Your dated entries + notes (append-only) โ”‚ โ”œโ”€โ”€ research/ # LLM-authored wiki pages โ”‚ โ””โ”€โ”€ raw/ # PDFs, HTML captures โ”œโ”€โ”€ logs/ # Chronological operation log โ”œโ”€โ”€ artifacts/ # Briefs, exports, lint findings โ””โ”€โ”€ state/builder/snapshots/ # context.md snapshots for rollback ``` No proprietary database โ€” what you see on disk is what MindBase knows. `git init` it, back it up with anything, delete a project with `rm -rf`. ## Architecture at a glance ``` Claude Code / Cursor / any MCP editor Web UI (npx mindbase-app) โ”‚ MCP ยท sub-agents with โ”‚ /commands ยท โ”‚ per-agent tool allowlists โ”‚ approval cards โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ฌโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ฌโ”€โ”€โ”€โ”€โ”€โ”€โ”˜ โ–ผ โ–ผ โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ” โ”‚ One ops engine: gather context โ†’ single โ”‚ โ”‚ constrained JSON completion โ†’ human โ”‚ โ”‚ approval โ†’ whitelisted executors โ†’ log โ”‚ โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ฌโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜ โ–ผ ~/mindbase-data/ ยท plain markdown (LLM: Ollama local models or any cloud key) ``` Monorepo: `packages/core` (TS strict library) ยท `apps/mcp` (49-tool MCP server) ยท `apps/server` + `apps/web` (Express + React UI) ยท `apps/app` (npx launcher) ยท `apps/plugin` (Claude Code bundle). ## Docs & help - **[Guide](https://github.com/frankchu91/mindbase/blob/HEAD/docs/guide.md)** โ€” first project, the four daily workflows, multi-project routing, troubleshooting - **[Website](https://frankchu91.github.io/mindbase-llm-wiki/)** โ€” live demos - **[CHANGELOG](https://github.com/frankchu91/mindbase/blob/HEAD/CHANGELOG.md)** ยท **[Issues](https://github.com/frankchu91/mindbase-llm-wiki/issues)** โ€” I reply to every issue same-day during beta - **The idea:** [Karpathy's LLM Wiki gist](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f) ## Roadmap **Next:** browser extension for one-click page capture ยท audio input via Whisper ยท unified meta-tool for Cursor/Windsurf slash-like UX. **Later:** team projects with human-in-the-loop review ยท audio digests ยท desktop app ยท mobile capture. ## Feedback Beta through 2026-Q4. If you tried MindBase and gave up โ€” please tell me why: [issues](https://github.com/frankchu91/mindbase-llm-wiki/issues) or [haobing0304@gmail.com](mailto:haobing0304@gmail.com). The blockers you hit are gold. ## License [MIT](https://github.com/frankchu91/mindbase/blob/HEAD/LICENSE) โ€” do what you want, no warranty. If you build something interesting on top, I'd love to hear about it. --- **Built with the belief that AI's most valuable gift is not "generation on demand" but "gardening of a persistent artifact you own."**