AI-powered second brain template for Claude Code + Obsidian

Capture insights, discover connections, and synthesize knowledge—with AI assistance.

Project Cornelius = Claude Code + Custom Agents + Obsidian + Switchable Knowledge Bases

It's like having a highly specialized AI research assistant that:

• 🧠 Manages multiple "brains" (knowledge vaults) you can switch between
• 🔍 Finds hidden connections in your notes you didn't know existed
• ✍️ Writes articles from your accumulated insights
• 🎯 Captures unique thoughts while preserving your voice
• 🔄 Evolves with you through Git-tracked configurations

One command to switch contexts:

/switch-brain /path/to/work-vault    # Work mode
/switch-brain /path/to/personal-vault # Personal mode
/switch-brain /path/to/research-vault # Research mode

Project Cornelius is a multi-layered knowledge management system that creates an intelligent bridge between your thinking and AI assistance. It's an agent-within-an-agent architecture that transforms Claude Code into a specialized second brain operator.

graph TB
    subgraph "User Space"
        User[User]
    end

    subgraph "Claude Code Layer"
        CC[Claude Code IDE]
        CLAUDE[CLAUDE.md System Prompt]
    end

    subgraph "Project Cornelius"
        CONFIG[.claude Config]
        AGENTS[Specialized Agents]
        COMMANDS[Slash Commands]
        MCP[MCP Servers]
    end

    subgraph "Knowledge Layer"
        BRAIN1[Brain 1 - Work]
        BRAIN2[Brain 2 - Personal]
        BRAIN3[Brain 3 - Research]
    end

    subgraph "Storage"
        GIT[Git Repository]
        OBS[Obsidian App]
    end

    User --> CC
    CC --> CLAUDE
    CLAUDE --> CONFIG
    CONFIG --> AGENTS
    CONFIG --> COMMANDS
    AGENTS --> MCP
    COMMANDS --> MCP
    MCP --> BRAIN1
    MCP -.->|switch| BRAIN2
    MCP -.->|switch| BRAIN3
    BRAIN1 --> GIT
    BRAIN1 --> OBS

    style BRAIN1 fill:#e1f5e1
    style BRAIN2 fill:#f0f0f0
    style BRAIN3 fill:#f0f0f0

graph LR
    subgraph "Input Layer"
        INPUT[User Request]
    end

    subgraph "Processing Pipeline"
        CLAUDE_PROMPT[System Prompt]
        ROUTER[Command Router]
        PROCESS[Processing Engine]
    end

    subgraph "Knowledge Operations"
        CAPTURE[Capture Insights]
        CONNECT[Connect Ideas]
        CREATE[Create Content]
        SEARCH[Search Knowledge]
    end

    subgraph "Brain Interface"
        VAULT[Active Brain]
        EMBED[Embeddings]
        GRAPH[Knowledge Graph]
    end

    subgraph "Output"
        NOTES[Permanent Notes]
        ARTICLES[Articles]
        CONNECTIONS[Discovered Links]
    end

    INPUT --> CLAUDE_PROMPT
    CLAUDE_PROMPT --> ROUTER
    ROUTER --> PROCESS

    PROCESS --> CAPTURE
    PROCESS --> CONNECT
    PROCESS --> CREATE
    PROCESS --> SEARCH

    CAPTURE --> VAULT
    CONNECT --> GRAPH
    CREATE --> VAULT
    SEARCH --> EMBED

    VAULT --> NOTES
    GRAPH --> CONNECTIONS
    VAULT --> ARTICLES

    style VAULT fill:#ffd700

mindmap
  root((Project Cornelius))
    Capture
      Extract Insights
      Preserve Voice
      Tag and Organize
      Source Attribution
    Connect
      Find Patterns
      Bridge Domains
      Surface Relationships
      Build Networks
    Create
      Synthesize Notes
      Generate Articles
      Build Frameworks
      Develop Ideas
    Evolve
      Track Changes
      Review History
      Discover Trends
      Refine Thinking

1. You interact with Claude Code normally
2. CLAUDE.md configures Claude as a specialized knowledge assistant
3. Agents perform specific knowledge tasks autonomously
4. Commands provide quick access to common operations
5. MCP Servers bridge to your Obsidian vault(s)
6. Brains (vaults) can be switched with a single command
7. Git tracks your agent configurations (not your private notes)

# Working on a project
/switch-brain /path/to/work-vault

# Personal knowledge work
/switch-brain /path/to/personal-vault

# Research mode
/switch-brain /path/to/research-vault

Each "brain" is a complete Obsidian vault with its own knowledge graph, allowing context switching between different domains of knowledge.

Project Cornelius creates a powerful nested agent system:

┌─────────────────────────────────────────┐
│         👤 Human (You)                  │
├─────────────────────────────────────────┤
│         🤖 Claude Code                  │ ← General AI assistant
├─────────────────────────────────────────┤
│     🧠 Project Cornelius Agent          │ ← Specialized for knowledge work
│     (Defined by CLAUDE.md)              │
├─────────────────────────────────────────┤
│     📚 Specialized Sub-Agents           │ ← Task-specific capabilities
│  (vault-manager, connection-finder...)  │
├─────────────────────────────────────────┤
│         💾 Your Knowledge Base          │ ← Your actual "brain"
│        (Obsidian Vault/Brain)           │
└─────────────────────────────────────────┘

1. Separation of Concerns

   • Code/config in Git (shareable, versionable)
   • Personal knowledge in Obsidian (private, searchable)
   • AI assistance via Claude Code (intelligent, contextual)

2. Hot-Swappable Brains

   • Work brain during office hours
   • Personal brain for life management
   • Research brain for deep dives
   • Creative brain for projects

3. Progressive Enhancement

   • Start with basic note capture
   • Add agents as needed
   • Customize prompts for your thinking style
   • Build your own commands

4. Knowledge Compounding

   • Every insight captured
   • Connections discovered automatically
   • Patterns emerge over time
   • Articles generated from your notes

# 1. Clone this repository
git clone https://github.com/vybe/project_cornelius.git
cd project_cornelius

# 2. Configure your vault path
cp .claude/settings.md.template .claude/settings.md
# Edit .claude/settings.md and set your vault path:
# VAULT_BASE_PATH=/path/to/your/vault

# 3. Configure MCP servers
# Edit .mcp.json and update all vault paths to match your vault
# Or use: /switch-brain /path/to/your/vault

# 4. Start Claude Code
claude

→ QUICKSTART.md for 5-minute setup → INSTALL.md for detailed installation → MCP-SETUP.md for MCP server configuration (required)

• vault-manager - Create and organize notes
• connection-finder - Find hidden relationships
• auto-discovery - Discover cross-domain patterns
• insight-extractor - Extract insights from content
• diagram-generator - Create visualizations

• /search-vault <query> - Quick search
• /recall <topic> - Deep 3-layer semantic search
• /find-connections <note> - Discover relationships
• /analyze-kb - Generate structure report
• /switch-brain <path> - Switch to different vault

Complete folder structure with templates:

• 00-Inbox/ - Capture zone
• 01-Sources/ - Books, articles, videos
• 02-Permanent/ - Your atomic insights (core)
• 03-MOCs/ - Navigation hubs
• 04-Output/ - Articles, frameworks, insights
• 05-Meta/ - Changelogs, templates

Capture: Extract insights from books and articles Connect: Find non-obvious relationships between ideas Create: Synthesize notes into articles and frameworks Evolve: Track how your thinking changes over time

Atomic notes - One idea per note Your words - Not copy-paste from sources Rich links - Connect everything Regular discovery - Find patterns with agents Active synthesis - Create from connections

• Claude Code
• Obsidian
• Node.js 18+
• MCP servers (via npm)

MIT - Use, modify, distribute freely. See LICENSE.

Questions? Check the docs above or start with QUICKSTART.md 🧠✨