Calibre Librarian MCP Server
Surfaces your Calibre e-book catalog to Claude via the Model Context Protocol, enabling search, metadata management, and library maintenance through natural language.
README
Calibre Librarian MCP Server
Model Context Protocol (MCP) server that surfaces your Calibre catalog to Claude via xmcp.
<!-- markdownlint-disable MD033 --> <p align="center"> <img src="assets/mcp-calibre.png" alt="Claude MCP view of Calibre Librarian" width="420" /> </p> <!-- markdownlint-enable MD033 -->
Requirements
- Node.js:
v24.13.0(auto-managed if you usenvm use). - pnpm: version
10.28.0or newer. - Calibre CLI tools:
calibredbandebook-convertmust be installable on your PATH. - Environment variables:
CALIBRE_LIBRARY_PATH– absolute path to your Calibre library directory.CALIBRE_DB_COMMAND– location of thecalibredbexecutable (e.g.,/opt/homebrew/bin/calibredb).FAVORITE_SEARCH_ENGINE_URL– base URL used when the server offers external book lookups (defaults to DuckDuckGo:https://duckduckgo.com/?q=).
Setup
-
Clone and enter the repo:
git clone https://github.com/chepetime/calibre-librarian-mcp.git cd calibre-librarian-mcp -
Install dependencies with pnpm:
pnpm install -
Copy the sample environment file and fill in your paths:
cp .env.example .env
Update the variables so the server can reach your Calibre library.
The env is just for local development. For Claude Desktop, you'll need to configure the server in the Claude Desktop settings.
Local development workflow
Use these scripts while iterating locally:
pnpm run dev– watches files and serves the MCP server over stdio.pnpm run lint– type-checks and lints the project.pnpm test– runs the full unit test suite once.pnpm test:watch– reruns tests whenever source files change.
Build for Claude Desktop (no Docker)
Follow this flow when you want Claude Desktop (or any MCP client) to run the compiled server directly:
-
Build the project so
dist/stdio.jsexists:pnpm run build -
(Optional) Run the built output locally for a quick smoke test:
pnpm start # equivalent to: node dist/stdio.js -
Configure Claude Desktop by editing
~/Library/Application Support/Claude/claude_desktop_config.json(or via the in-app UI). Setcommandtonode, include the absolute path todist/stdio.jsas the firstargsentry, and provide the required environment variables:
{
"globalShortcut": "",
"mcpServers": {
"calibre-librarian": {
"command": "node",
"args": ["/Users/you/path/to/calibre-librarian-mcp/dist/stdio.js"],
"env": {
"CALIBRE_LIBRARY_PATH": "<Absolute path to your>/Calibre",
"CALIBRE_DB_COMMAND": "/opt/homebrew/bin/calibredb",
"FAVORITE_SEARCH_ENGINE_URL": "https://duckduckgo.com/?q="
}
}
},
"preferences": {
"quickEntryShortcut": "off",
"menuBarEnabled": false
}
}
After Claude Desktop reloads, it will list calibre-librarian as an available MCP server whenever MCP-enabled conversations start.
Docker deployment
Run the server in a container with Calibre pre-installed when you prefer an isolated environment.
Quick start
# Build the image
docker build -t calibre-librarian-mcp .
# Run with your Calibre library mounted
docker run -it \
-v /path/to/your/calibre/library:/library:ro \
-e CALIBRE_LIBRARY_PATH=/library \
calibre-librarian-mcp
Docker Compose
-
Copy and customize
docker-compose.yml. -
Set your library path and launch the stack:
export CALIBRE_LIBRARY_PATH=/path/to/your/calibre/library docker compose up --build
Claude Desktop with Docker
To let Claude Desktop run the container directly, point it at docker run:
{
"mcpServers": {
"calibre-librarian": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-v",
"/path/to/your/calibre/library:/library:ro",
"-e",
"CALIBRE_LIBRARY_PATH=/library",
"calibre-librarian-mcp"
]
}
}
}
Note: Remove :ro from the volume mount and add -e CALIBRE_ENABLE_WRITE_OPERATIONS=true to enable write operations.
Tool Catalog & Examples
Installing the MCP CLI
The examples below use the mcp CLI published by Anthropic. Install (or run) it with any of the following options:
-
Global install (recommended if you call MCP tools frequently):
npm install -g @anthropic-ai/mcp-cli # now `mcp --help` should work -
One-off execution without a global install:
npx @anthropic-ai/mcp-cli --help # or pnpm dlx @anthropic-ai/mcp-cli --help
CLI usage
All tools can be invoked from MCP Inspector or the CLI:
mcp call calibre-librarian <toolName> '<json payload>'
Prompts (e.g., merge_duplicates, library_cleanup, search_library) use the companion command:
mcp prompt calibre-librarian library_cleanup '{"focus":"missing covers"}'
Swap library_cleanup for any prompt listed below, and replace calibre-librarian with the server name you configured in mcp.json.
Library Overview & Metadata
| Tool | Example |
|---|---|
list_sample_books |
mcp call calibre-librarian list_sample_books '{"limit":5}' |
get_book_details |
mcp call calibre-librarian get_book_details '{"bookId":42}' |
get_library_stats |
mcp call calibre-librarian get_library_stats '{}' |
get_all_tags |
mcp call calibre-librarian get_all_tags '{"sortBy":"count","minCount":5}' |
get_custom_columns |
mcp call calibre-librarian get_custom_columns '{"includeDisplay":true}' |
Search & Discovery
| Tool | Example |
|---|---|
search_books |
mcp call calibre-librarian search_books '{"query":"author:Sanderson and tag:fantasy","limit":10}' |
search_books_by_title |
mcp call calibre-librarian search_books_by_title '{"title":"stormlight","exact":false}' |
search_authors_by_name |
mcp call calibre-librarian search_authors_by_name '{"name":"ng","sortBy":"count"}' |
get_books_by_author |
mcp call calibre-librarian get_books_by_author '{"author":"Robin Hobb","sortBy":"series","ascending":true}' |
get_books_by_author_id |
mcp call calibre-librarian get_books_by_author_id '{"authorId":17}' |
get_books_by_series |
mcp call calibre-librarian get_books_by_series '{"series":"The Expanse","exact":true}' |
get_books_by_tag |
mcp call calibre-librarian get_books_by_tag '{"tag":"cozy mystery","limit":25}' |
search_books_by_tag_pattern |
mcp call calibre-librarian search_books_by_tag_pattern '{"pattern":"*punk","limit":10}' |
Full-Text & Content Access
| Tool | Example |
|---|---|
full_text_search |
mcp call calibre-librarian full_text_search '{"query":"\"winter is coming\"","matchAll":false}' |
search_book_content |
mcp call calibre-librarian search_book_content '{"bookId":12,"query":"quantum","contextChars":120}' |
fetch_excerpt |
mcp call calibre-librarian fetch_excerpt '{"bookId":8,"maxChars":1500}' |
Cleanup & Duplicate Workbench
| Tool | Example |
|---|---|
find_duplicates |
mcp call calibre-librarian find_duplicates '{"mode":"author_title","threshold":0.85}' |
compare_books |
mcp call calibre-librarian compare_books '{"bookIds":[101,205],"fields":["title","series","formats"]}' |
quality_report |
mcp call calibre-librarian quality_report '{"checks":["missing_cover","missing_tags"],"limit":20}' |
merge_duplicates prompt |
mcp prompt calibre-librarian merge_duplicates '{"bookIds":[101,205]}' |
library_cleanup prompt |
mcp prompt calibre-librarian library_cleanup '{"focus":"missing covers"}' |
search_library prompt |
mcp prompt calibre-librarian search_library '{"query":"hopepunk","searchType":"tag"}' |
Smart Maintenance Recipes
| Tool | Example |
|---|---|
normalize_author_sort |
mcp call calibre-librarian normalize_author_sort '{"preview":true,"limit":25}' |
bulk_retag |
mcp call calibre-librarian bulk_retag '{"query":"author:Sanderson","action":"add","tags":"cosmere","preview":true}' |
library_maintenance |
mcp call calibre-librarian library_maintenance '{"operation":"check"}' |
missing_book_scout |
mcp call calibre-librarian missing_book_scout '{"readingList":"Dune\n1984\nThe Hobbit","searchEngine":"annas_archive"}' |
Metadata Editing & Custom Columns
Requires
CALIBRE_ENABLE_WRITE_OPERATIONS=true
| Tool | Example |
|---|---|
set_custom_column |
mcp call calibre-librarian set_custom_column '{"bookId":42,"column":"#reading_status","value":"Started"}' |
set_metadata |
mcp call calibre-librarian set_metadata '{"bookId":42,"title":"The Final Empire (Revised)","tags":["cosmere","favorite"]}' |
Setup & Configuration Tools
| Tool | Example |
|---|---|
generate_claude_config |
mcp call calibre-librarian generate_claude_config '{"enableWrites":false}' |
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
CALIBRE_LIBRARY_PATH |
Yes | — | Absolute path to your Calibre library directory |
CALIBRE_DB_COMMAND |
No | calibredb |
Path to the calibredb executable |
CALIBRE_COMMAND_TIMEOUT_MS |
No | 15000 |
Timeout for calibredb commands in milliseconds |
CALIBRE_ENABLE_WRITE_OPERATIONS |
No | false |
Enable metadata editing tools (set_metadata, etc.) |
FAVORITE_SEARCH_ENGINE_URL |
No | https://duckduckgo.com/?q= |
Base URL for external book search links |
MCP_SERVER_NAME |
No | Calibre Librarian MCP |
Server name shown in MCP clients |
Resources
The server exposes these MCP resources:
| URI | Description |
|---|---|
calibre://library/info |
Library configuration and statistics |
calibre://library/custom-columns |
Custom column definitions |
calibre://docs/inspector-guide |
MCP Inspector verification guide |
Troubleshooting
"calibredb: command not found"
The server can't find the Calibre CLI tools. Solutions:
- macOS (Homebrew):
brew install calibreor setCALIBRE_DB_COMMAND=/Applications/calibre.app/Contents/MacOS/calibredb - macOS (App):
CALIBRE_DB_COMMAND=/Applications/calibre.app/Contents/MacOS/calibredb - Windows:
CALIBRE_DB_COMMAND=C:\Program Files\Calibre2\calibredb.exe - Linux: Install Calibre via package manager, usually adds
calibredbto PATH
"Library path does not exist"
Verify your CALIBRE_LIBRARY_PATH:
# Check the path contains metadata.db
ls "$CALIBRE_LIBRARY_PATH/metadata.db"
"Write operations are disabled"
Write tools (set_metadata, set_custom_column, bulk_retag with preview:false, etc.) require:
CALIBRE_ENABLE_WRITE_OPERATIONS=true
Add this to your .env file or Claude Desktop config.
"Command timed out"
For large libraries, increase the timeout:
CALIBRE_COMMAND_TIMEOUT_MS=60000 # 60 seconds
"Full-text search returns no results"
Calibre FTS must be enabled:
- Open Calibre
- Go to Preferences → Searching
- Enable Full text searching
- Click Re-index all books
Server not appearing in Claude Desktop
-
Verify the config file path:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json - Linux:
~/.config/Claude/claude_desktop_config.json
- macOS:
-
Check JSON syntax is valid
-
Restart Claude Desktop completely
-
Check Claude Desktop logs for errors
Testing with MCP Inspector
Use the built-in verification guide:
# Start dev server
npm run dev
# In another terminal, run inspector
npx @anthropic/mcp-inspector
Or use the generate_claude_config tool to get your configuration.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.