Notidium

Your developer knowledge, elementally organized.

A developer-focused, local-first note-taking application with semantic search and native MCP integration.

Philosophy

• Context is everything — Code without context is just syntax. Notidium preserves the why, not just the what.
• Search, don't organize — Folders are a tax on future you. Semantic search means you write notes, not maintain systems.
• Local-first, AI-native — Your knowledge stays on your machine. AI assistants access it through MCP, not cloud APIs.
• Developer-specific by design — Not a general PKM adapted for code. Built from the ground up for how developers think and work.

Features

• Full-text search powered by Tantivy
• Semantic search with BGE-small-en embeddings (384 dimensions)
• MCP server for Claude Desktop and other AI assistants
• Markdown-first with YAML frontmatter support
• File-based storage — your notes are just .md files
• HTTP API for web UI integration

Installation

# Install from crates.io
cargo install notidium

# Or build from source
git clone https://github.com/pjankiewicz/notidium
cd notidium
cargo install --path .

Quick Start

# Initialize a new vault
notidium init ~/Notidium

# Add some notes to ~/Notidium/notes/
echo "# My First Note\n\nHello, Notidium!" > ~/Notidium/notes/hello.md

# Index all notes (builds full-text + semantic search)
notidium index

# Search your notes
notidium search "hello"
notidium search --semantic "greeting message"

# Start the server (HTTP API + MCP + Web UI)
notidium serve

# Start MCP-only server for Claude Desktop (stdio mode)
notidium mcp

CLI Commands

MCP Integration

Notidium includes a built-in MCP server that exposes your knowledge base to AI assistants like Claude.

Claude Desktop Setup

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "notidium": {
      "command": "/path/to/notidium",
      "args": ["mcp"]
    }
  }
}

Available MCP Tools

HTTP API

When running notidium serve, the following endpoints are available:

GET  /health                 Health check
GET  /api/notes              List notes
POST /api/notes              Create note
GET  /api/notes/:id          Get note by ID
PUT  /api/notes/:id          Update note
DELETE /api/notes/:id        Delete note
GET  /api/search             Full-text search (?q=query)
GET  /api/search/semantic    Semantic search (?q=query)
GET  /api/notes/:id/related  Find related notes
POST /api/capture            Quick capture
GET  /api/tags               List all tags
GET  /api/stats              Get statistics
GET  /api/docs               Swagger UI
GET  /api/openapi.json       OpenAPI spec
POST /mcp                    MCP protocol endpoint

Vault Structure

~/Notidium/
├── notes/                   # All markdown files
│   ├── inbox/               # Quick captures land here
│   └── *.md                 # Your notes
├── attachments/             # Images, PDFs, etc.
├── templates/               # Note templates
├── .notidium/               # App data (gitignore this)
│   ├── config.toml          # User preferences
│   ├── index.db             # SQLite metadata
│   ├── tantivy/             # Full-text search index
│   ├── vectors/             # Vector embeddings (LanceDB)
│   ├── chunks.json          # Embedded chunks cache
│   └── logs/                # Debug logs
└── .notidiumignore          # Files to skip indexing

Note Format

Notes are standard Markdown files with optional YAML frontmatter:

---
tags: [rust, oauth, tutorial]
aliases: [OAuth Setup, Auth Guide]
---

# Setting up OAuth in Rust

Here's how I configured OAuth for my Axum app...

```rust
let oauth_client = BasicClient::new(
    ClientId::new(client_id),
    Some(ClientSecret::new(client_secret)),
    AuthUrl::new(auth_url)?,
    Some(TokenUrl::new(token_url)?),
);

The key insight is to store tokens in Redis...

## Configuration

Create `~/Notidium/.notidium/config.toml`:

```toml
# Vault settings
notes_dir = "notes"
attachments_dir = "attachments"
templates_dir = "templates"

# Server settings
http_port = 3939
mcp_port = 3940

# Embedding settings
[embedding]
prose_model = "BAAI/bge-small-en-v1.5"
batch_size = 32

# Search settings
[search]
default_limit = 10
max_limit = 100

Tech Stack

• Rust — Fast, safe, no runtime
• Axum — Async HTTP framework
• Tantivy — Full-text search engine
• fastembed — Local embedding inference
• LanceDB — Vector database
• SQLite — Metadata storage
• rmcp — MCP protocol implementation
• pulldown-cmark — Markdown parsing

Development

# Run in development
cargo run -- serve

# Run tests
cargo test

# Build release
cargo build --release

Contributing

Contributions are welcome. Please open an issue first to discuss what you would like to change.

License

MIT

Notidium — The element of context.