DataBook

An AI-editable SQL database app powered by MCP. The MCP server is the app — its tool definitions encode the product decisions. The AI's capabilities become the feature set.

What is DataBook?

DataBook is a Model Context Protocol (MCP) server that gives AI assistants full control over a collection of SQLite databases. Instead of building a traditional CRUD UI, DataBook exposes database operations as MCP tools — letting any MCP-capable AI client (Claude, Cursor, etc.) act as the interface.

Tools

Database Management

Tool	Description
list_databases	List all available databases
select_database	Choose which database to work with
create_database	Create a new empty database
delete_database	Delete a database permanently

Schema Management

Tool	Description
list_tables	List all tables in the selected database
describe_table	Show table schema and column definitions
create_table	Create a table with column definitions
alter_table	Add, rename, drop columns, or rename the table
drop_table	Delete a table and all its data

Data CRUD

Tool	Description
insert_rows	Insert one or more rows
update_rows	Update rows matching a WHERE condition
delete_rows	Delete rows matching a WHERE condition

Query

Tool	Description
query	Execute any SQL statement and return results

Query Pages (Views)

Tool	Description
create_query_page	Save a named query with a view type (table, kanban, calendar, etc.)
list_query_pages	List all saved query pages for the current database
update_query_page	Rename, change query, or change view type
delete_query_page	Remove a saved query page

Storage

• Each database is a separate .db SQLite file stored in ~/.databook/ (configurable via DATABOOK_DIR)
• Query pages are stored in ~/.databook/.meta.db — a metadata SQLite database

Setup

Install

npm install
npm run build

Run

npm start
# or
node dist/index.js

Configure data directory

DATABOOK_DIR=/path/to/data node dist/index.js

Connecting to Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "databook": {
      "command": "node",
      "args": ["/path/to/DataBook/dist/index.js"],
      "env": {
        "DATABOOK_DIR": "/path/to/your/data"
      }
    }
  }
}

Or using npx after publishing:

{
  "mcpServers": {
    "databook": {
      "command": "npx",
      "args": ["databook"]
    }
  }
}

Architecture

DataBook MCP Server (TypeScript)
  │
  ├── DatabaseManager          # Core logic layer
  │   ├── SQLite files         # One .db file per database (via better-sqlite3)
  │   └── .meta.db             # Query pages metadata
  │
  └── McpServer (stdio)        # MCP transport
      └── 17 tools             # The complete feature set

Development

npm run dev    # Watch mode (recompiles on change)
npm run build  # Production build