🧠 DeepBrain

Personal AI brain. Store, search, and grow your knowledge.

DeepBrain is a personal knowledge engine inspired by GBrain, rebuilt from scratch with:

• 🔌 Multi-provider — 7 embedding providers via agentkits
• 🇨🇳 Chinese-friendly — DeepSeek, DashScope (Qwen), Zhipu (GLM) built-in
• 🏠 Local-first — Ollama + PGLite, zero cloud dependency
• ⚡ Lightweight — No bloat, just the essentials
• 🔍 Hybrid search — Vector (HNSW) + keyword (tsvector) + RRF fusion
• 📊 Compiled Truth + Timeline — Current facts vs. historical events
• 🔗 Knowledge graph — Bidirectional links between pages
• 💤 Dream Cycle — Automated knowledge maintenance

Quick Start

# Install
npm install -g deepbrain

# Initialize with your preferred provider
deepbrain init ollama          # Local, free
deepbrain init deepseek        # Cheap, fast
deepbrain init gemini          # Free tier available

# Add knowledge
deepbrain put meeting-notes notes.md
deepbrain put "john-doe" "John is a senior engineer at Google. Expert in distributed systems."

# Search
deepbrain query "who knows about distributed systems?"
deepbrain search "Google engineer"

# Link knowledge
deepbrain link john-doe google-company

# Timeline
deepbrain timeline john-doe "Promoted to Staff Engineer"

# Stats & maintenance
deepbrain stats
deepbrain dream

Programmatic API

import { Brain } from 'deepbrain';

const brain = new Brain({
  embedding_provider: 'deepseek',
  api_key: process.env.DEEPSEEK_API_KEY,
});
await brain.connect();

// Store
await brain.put('ray', {
  type: 'person',
  title: 'Ray Wang',
  compiled_truth: 'CEO of Deepleaper. Expert in AI agents and context intelligence.',
});

// Semantic search
const results = await brain.query('who builds AI agents?');
console.log(results[0].slug); // → 'ray'

// Link
await brain.link('ray', 'deepleaper', 'founder of');

// Timeline
await brain.addTimeline('ray', {
  date: '2026-04-13',
  summary: 'Started DeepBrain project',
});

// Stats
const stats = await brain.stats();
console.log(stats);

await brain.disconnect();

Providers

How It Works

Your Knowledge → Pages (Markdown) → Chunks → Embeddings → Vector DB
                    ↓                                        ↓
                 Links (Graph)                         Hybrid Search
                    ↓                                   (HNSW + tsvector + RRF)
                Timeline (append-only)                     ↓
                    ↓                                  Results
              Dream Cycle (auto-maintenance)

Core Concepts

vs. GBrain

Architecture

deepbrain
├── src/
│   ├── index.ts         # Main exports
│   ├── cli.ts           # CLI (deepbrain put/get/query/...)
│   ├── core/
│   │   ├── brain.ts     # Brain engine (CRUD + search + graph)
│   │   ├── schema.ts    # PGLite schema (pages, chunks, links, tags, timeline)
│   │   └── types.ts     # TypeScript types
│   └── dream/
│       └── index.ts     # Dream Cycle (auto-maintenance)
├── package.json
└── tsconfig.json

Dependencies: Only @electric-sql/pglite + agentkits (which wraps openai SDK).

Roadmap

• Core engine (CRUD, search, graph, timeline)
• Hybrid search (vector + keyword + RRF)
• CLI
• Dream Cycle v1
• Multi-provider embedding (via agentkits)
• MCP Server (for Claude/Cursor/OpenClaw integration)
• Web UI (knowledge graph visualization)
• Import from Notion / Obsidian / WeChat
• MRG layer (mission-centric knowledge graph)
• Smart chunking (semantic + LLM-assisted)
• Multi-brain sync

License

MIT © Magicray1217

Inspired by GBrain by Garry Tan. Built with AgentKits.