Skip to content

API.md

Latest commit

 

History

History
698 lines (511 loc) · 21.4 KB

API.md

File metadata and controls

698 lines (511 loc) · 21.4 KB
Raw

CalibreMCP API Reference

Comprehensive reference for all MCP tools and endpoints

🔧 MCP Tools

CalibreMCP uses FastMCP 2.13+ with portmanteau tools - consolidated tools that handle multiple related operations through an operation parameter. This reduces tool count while maintaining full functionality.

Tool 1: query_books ⭐ PRIMARY BOOK ACCESS TOOL

Portmanteau tool that consolidates all book querying operations: search, list, find, get books by various criteria.

Verb Mapping for Claude

IMPORTANT: Users may use different verbs (search, list, find, query, get) but they all map to the same operation.

ALL of these user requests should use operation="search":

  • "search books by [author]" → query_books(operation="search", author="...")
  • "list books by [author]" → query_books(operation="search", author="...")
  • "find books by [author]" → query_books(operation="search", author="...")
  • "query books by [author]" → query_books(operation="search", author="...")
  • "get books by [author]" → query_books(operation="search", author="...")
  • "show me books by [author]" → query_books(operation="search", author="...")

Rule: If the user wants to access/retrieve/discover books with filters (author, tag, publisher, etc.), use operation="search" regardless of which verb they use.

Signature

async def query_books(
    operation: str,  # "search", "list", "by_author", "by_series"
    # Search parameters (for operation="search")
    author: Optional[str] = None,
    authors: Optional[List[str]] = None,
    exclude_authors: Optional[List[str]] = None,
    text: Optional[str] = None,
    query: Optional[str] = None,  # Alias for text
    tag: Optional[str] = None,
    tags: Optional[List[str]] = None,
    exclude_tags: Optional[List[str]] = None,
    series: Optional[str] = None,
    exclude_series: Optional[List[str]] = None,
    publisher: Optional[str] = None,
    publishers: Optional[List[str]] = None,
    has_publisher: Optional[bool] = None,
    rating: Optional[int] = None,
    min_rating: Optional[int] = None,
    max_rating: Optional[int] = None,
    unrated: Optional[bool] = None,
    pubdate_start: Optional[str] = None,
    pubdate_end: Optional[str] = None,
    added_after: Optional[str] = None,
    added_before: Optional[str] = None,
    min_size: Optional[int] = None,
    max_size: Optional[int] = None,
    formats: Optional[List[str]] = None,
    comment: Optional[str] = None,
    has_empty_comments: Optional[bool] = None,
    format_table: bool = False,
    limit: int = 50,
    offset: int = 0,
    # For operation="by_author"
    author_id: Optional[int] = None,
    # For operation="by_series"
    series_id: Optional[int] = None,
    # For operation="list"
    sort: str = "title",
) -> Dict[str, Any]

Operations

  1. operation="search" ⭐ PRIMARY OPERATION

    • Use for ALL user requests to access books with filters
    • Works with ANY verb: "search", "list", "find", "query", "get", "show me"
    • Supports comprehensive filtering (author, publisher, year, tags, rating, etc.)
    • Example: query_books(operation="search", author="Conan Doyle")

  2. operation="list"

    • Only use when user explicitly wants ALL books without filters
    • Simple pagination with minimal filtering
    • Example: query_books(operation="list", limit=20)

  3. operation="by_author"

    • Get books by numeric author_id (requires author_id parameter)
    • For author names, use operation="search" with author parameter instead
    • Example: query_books(operation="by_author", author_id=42)

  4. operation="by_series"

    • Get books in a series by numeric series_id (requires series_id parameter)
    • For series names, use operation="search" with series parameter instead
    • Example: query_books(operation="by_series", series_id=15)

Search Parameters (for operation="search")

Author Filters:

  • author: Filter by author name (case-insensitive partial match)
  • authors: Filter by multiple authors (OR logic)
  • exclude_authors: Exclude books by these authors

Text & Tags:

  • text / query: Search in title, author, tags, series, comments
  • tag: Filter by single tag
  • tags: Filter by multiple tags (OR logic)
  • exclude_tags: Exclude books with these tags

Series:

  • series: Filter by series name
  • exclude_series: Exclude books in these series

Rating:

  • rating: Exact rating (1-5)
  • min_rating: Minimum rating (1-5)
  • max_rating: Maximum rating (1-5)
  • unrated: Filter for unrated books only

Publisher:

  • publisher: Filter by publisher name
  • publishers: Filter by multiple publishers
  • has_publisher: Filter books with/without publisher

Date Filters:

  • pubdate_start: Publication date start (YYYY-MM-DD)
  • pubdate_end: Publication date end (YYYY-MM-DD)
  • added_after: Date added after (YYYY-MM-DD)
  • added_before: Date added before (YYYY-MM-DD)

File Filters:

  • min_size: Minimum file size in bytes
  • max_size: Maximum file size in bytes
  • formats: List of formats to include (e.g., ["EPUB", "PDF"])

Other:

  • comment: Search in book comments only
  • has_empty_comments: Filter books with empty/non-empty comments
  • format_table: Format results as pretty text table (boolean)

Pagination:

  • limit: Maximum results to return (default: 50)
  • offset: Results offset for pagination (default: 0)

Usage Examples

# User says: "list books by conan doyle"
result = await query_books(operation="search", author="Conan Doyle", format_table=True)

# Advanced search: author + publisher + year
result = await query_books(
    operation="search", author="Conan Doyle", publisher="Penguin", pubdate_start="1900-01-01", pubdate_end="1930-12-31"
)

# Search with multiple filters (AND logic)
result = await query_books(
    operation="search", author="Agatha Christie", tags=["mystery", "crime"], min_rating=4, exclude_tags=["horror"]
)

# Search by publisher with year range
result = await query_books(
    operation="search",
    publishers=["O'Reilly Media", "No Starch Press"],
    pubdate_start="2023-01-01",
    pubdate_end="2024-12-31",
)

# Search recently added books with rating
result = await query_books(operation="search", added_after="2024-01-01", min_rating=4, format_table=True)

# List all books (simple operation)
result = await query_books(operation="list", limit=20)

# Get books by author ID
result = await query_books(operation="by_author", author_id=42, limit=10)

# Get books in a series
result = await query_books(operation="by_series", series_id=15)

Returns

Dictionary containing operation-specific results:

  • For operation="search": Paginated search results with total count, books list, optional table
  • For operation="list": Simple book list with pagination
  • For operation="by_author": List of books by author ID
  • For operation="by_series": List of books in series order

Tool 2: manage_books ⭐ BOOK CRUD OPERATIONS

Portmanteau tool that consolidates book management operations: add, get, update, delete.

Signature

async def manage_books(
    operation: str,  # "add", "get", "update", "delete"
    book_id: Optional[str] = None,
    file_path: Optional[str] = None,
    metadata: Optional[Dict[str, Any]] = None,
    # ... additional parameters based on operation
) -> Dict[str, Any]

Operations

  1. operation="add": Add a new book to the library (requires file_path)
  2. operation="get": Retrieve detailed book information (requires book_id)
  3. operation="update": Update book metadata and properties (requires book_id)
  4. operation="delete": Delete a book from the library (requires book_id)

Usage Examples

# Add a book
result = await manage_books(
    operation="add", file_path="/path/to/book.epub", metadata={"title": "My Book", "authors": ["Author Name"]}
)

# Get book details
result = await manage_books(operation="get", book_id="123", include_metadata=True, include_formats=True)

# Update book metadata
result = await manage_books(operation="update", book_id="123", metadata={"rating": 5, "tags": ["favorite"]})

# Delete a book
result = await manage_books(operation="delete", book_id="123", delete_files=True)

Tool 3: manage_libraries ⭐ LIBRARY MANAGEMENT

Portmanteau tool for managing multiple Calibre libraries.

Operations

  1. operation="list": List all available libraries
  2. operation="switch": Switch to a different library (requires library_name)
  3. operation="stats": Get statistics for a library
  4. operation="search": Search across multiple libraries

Usage Examples

# List all libraries
result = await manage_libraries(operation="list")

# Switch library
result = await manage_libraries(operation="switch", library_name="Main Library")

# Get library statistics
result = await manage_libraries(operation="stats", library_name="Main Library")

# Search across libraries
result = await manage_libraries(
    operation="search", query="python programming", libraries=["Main Library", "IT Library"]
)

Tool 4: get_book_details

Get complete metadata and file information for a specific book (standalone tool, not portmanteau).

Signature

async def get_book_details(book_id: int) -> BookDetailResponse

Parameters

  • book_id (int): Calibre book ID to fetch details for

Returns

BookDetailResponse with complete book information:

  • Basic metadata: title, authors, series, rating, tags
  • Publication info: published date, languages, comments
  • File information: available formats, download links
  • System data: identifiers, last modified, cover URL

Tool 5: test_calibre_connection

Test connection to Calibre server and get diagnostics.

Signature

async def test_calibre_connection() -> ConnectionTestResponse

Returns

ConnectionTestResponse with:

  • connected: Boolean connection status
  • server_version: Calibre server version
  • library_name: Primary library name
  • total_books: Number of books in library
  • response_time_ms: Connection response time
  • error_message: Error details if connection failed
  • server_capabilities: List of supported features

Tool 6: manage_viewer ⭐ BOOK VIEWER OPERATIONS

Portmanteau tool for book viewer and file opening operations.

Operations

  1. operation="open": Open a book in the viewer (requires book_id, file_path)
  2. operation="get_page": Get a specific page from a book (requires book_id, file_path, page_number)
  3. operation="get_metadata": Get comprehensive metadata for a book (requires book_id, file_path)
  4. operation="get_state": Get current viewer state (requires book_id, file_path)
  5. operation="update_state": Update viewer state (requires book_id, file_path)
  6. operation="close": Close a viewer session (requires book_id)
  7. operation="open_file": Open book file with system's default application (requires book_id, file_path)
  8. operation="open_random" ⭐ NEW: Search for books matching criteria, randomly select one, and open it

open_random Operation

Searches for books matching author/tag/series filters, randomly selects one, and opens it with the system's default application.

Parameters:

  • author (optional): Author name filter (e.g., "John Dickson Carr")
  • tag (optional): Tag name filter (e.g., "mystery")
  • series (optional): Series name filter (e.g., "Sherlock Holmes")
  • format_preference (default: "EPUB"): Preferred file format

Note: At least one search filter (author, tag, or series) is required.

Returns:

{"success": bool, "book_id": int, "title": str, "author": str, "file_path": str, "format": str, "message": str}

Usage Examples:

# Open a random book by author
result = await manage_viewer(operation="open_random", author="Dickson Carr")

# Open random mystery book
result = await manage_viewer(operation="open_random", tag="mystery")

# Open random book in series with PDF preference
result = await manage_viewer(operation="open_random", series="Sherlock Holmes", format_preference="PDF")

Tool 7: manage_metadata ⭐ METADATA OPERATIONS

Portmanteau tool for metadata management and display.

Operations

  1. operation="update": Update metadata for single or multiple books (requires updates)
  2. operation="organize_tags": AI-powered tag organization and cleanup suggestions
  3. operation="fix_issues": Automatically fix common metadata problems
  4. operation="show" ⭐ NEW: Display comprehensive book metadata in formatted popup/modal

show Operation

Searches for a book by title or author, retrieves comprehensive metadata, and displays it in a formatted HTML popup (optional).

Parameters:

  • query (required): Search query (title or partial title) - e.g., "Gormenghast"
  • author (optional): Author name filter - e.g., "Mervyn Peake"
  • open_browser (default: True): Whether to open metadata in browser popup

Returns:

{
    "success": bool,
    "book_id": int,
    "title": str,
    "author": str,
    "metadata": dict,  # Complete metadata dictionary
    "html_path": Optional[str],  # Path to HTML file if open_browser=True
    "formatted_text": str,  # Formatted text representation
}

Usage Examples:

# Show metadata for a book (opens HTML popup)
result = await manage_metadata(operation="show", query="Gormenghast")

# Show metadata by author
result = await manage_metadata(operation="show", author="Peake")

# Show metadata without browser popup
result = await manage_metadata(operation="show", query="Gormenghast", open_browser=False)

# Display formatted text
if result["success"]:
    print(result["formatted_text"])

Additional Portmanteau Tools

  • manage_smart_collections: Create, update, delete, query smart collections
  • manage_users: User management and authentication operations
  • Export tools: export_books_csv, export_books_json, export_books_html, export_books_pandoc

📊 Response Models

BookSearchResult

Individual book result from search operations.

class BookSearchResult(BaseModel):
    book_id: int
    title: str
    authors: List[str]
    series: Optional[str] = None
    series_index: Optional[float] = None
    rating: Optional[int] = None  # 1-5 scale
    tags: List[str] = []
    languages: List[str] = ["en"]
    formats: List[str] = []  # ["EPUB", "PDF", "MOBI", etc.]
    last_modified: Optional[str] = None
    cover_url: Optional[str] = None

LibrarySearchResponse

Response from query_books search operations.

class LibrarySearchResponse(BaseModel):
    results: List[BookSearchResult]
    total_found: int
    query_used: Optional[str] = None
    search_time_ms: int

BookDetailResponse

Complete book information from get_book_details.

class BookDetailResponse(BaseModel):
    book_id: int
    title: str
    authors: List[str]
    series: Optional[str] = None
    series_index: Optional[float] = None
    rating: Optional[int] = None
    tags: List[str] = []
    comments: Optional[str] = None
    published: Optional[str] = None
    languages: List[str] = []
    formats: List[str] = []
    identifiers: Dict[str, str] = {}  # ISBN, Goodreads, etc.
    last_modified: Optional[str] = None
    cover_url: Optional[str] = None
    download_links: Dict[str, str] = {}  # Format -> URL mapping

ConnectionTestResponse

Server connection test results.

class ConnectionTestResponse(BaseModel):
    connected: bool
    server_version: Optional[str] = None
    library_name: Optional[str] = None
    total_books: Optional[int] = None
    response_time_ms: int
    error_message: Optional[str] = None
    server_capabilities: List[str] = []

🎯 Default Library Auto-Loading

IMPORTANT: The default library is automatically loaded on server startup. Users don't need to manually load or switch libraries before querying books.

Priority order for auto-loading:

  1. Persisted library (from FastMCP 2.13 storage)
  2. config.local_library_path (if set)
  3. Active library from Calibre's config files
  4. First discovered library (fallback)

🔍 Calibre Server REST API

CalibreMCP communicates with Calibre's built-in REST API server.

Base URL Structure

http://localhost:8080/ajax/...

Core Endpoints Used

  • GET /ajax/interface-data/init - Server initialization and library info
  • GET /ajax/search - Library search with query parameters
  • GET /ajax/books - Bulk book metadata retrieval
  • GET /ajax/book/{id} - Individual book details

Media Endpoints

  • GET /get/cover/{id} - Book cover image
  • GET /get/{format}/{id} - Download book in specific format
  • GET /get/thumb/{id} - Thumbnail cover image

Authentication

Supports HTTP Basic Authentication when enabled in Calibre server:

# Configuration
username: "your_username"
password: "your_password"

📝 Query Syntax

Search Query Patterns

CalibreMCP supports Calibre's native search syntax through the query_books tool.

Field-Specific Searches

# Use query_books with operation="search"
query_books(operation="search", author="Conan Doyle")  # Author filter
query_books(operation="search", tag="science")  # Tag filter
query_books(operation="search", series="Foundation")  # Series filter
query_books(operation="search", publisher="O'Reilly")  # Publisher filter
query_books(operation="search", min_rating=4)  # Rating filter

Boolean Operations

Filters in query_books use AND logic when combined:

  • author="X" AND tag="Y" AND min_rating=4 = All conditions must match
  • tags=["X", "Y"] = Books with tag X OR tag Y

Advanced Patterns

# Date range filtering
query_books(operation="search", pubdate_start="2020-01-01", pubdate_end="2024-12-31")

# File size filtering
query_books(
    operation="search",
    min_size=1048576,  # 1 MB
    max_size=10485760,  # 10 MB
)

# Format filtering
query_books(operation="search", formats=["EPUB", "PDF"])

Sort Options (for operation="list")

  • title - Alphabetical by title
  • authors - Alphabetical by author surname
  • rating - Highest rated first
  • date - Most recently added first
  • pubdate - Most recently published first
  • series - Series order

⚡ Performance Guidelines

Optimal Query Patterns

  • Use specific filters (author, tag) instead of broad text searches
  • Limit results to 50 or fewer for responsive UI
  • Use format_table=True for human-readable output
  • Cache frequent queries on client side

Rate Limiting

  • Default: 60 requests per minute
  • Burst limit: 10 concurrent requests
  • Timeout: 30 seconds per request

Memory Considerations

  • Cover images: ~100KB each
  • Book metadata: ~2KB each
  • Search results: Cache for 5 minutes
  • Total cache: Limited to 50MB

🚨 Error Codes

HTTP Status Codes

  • 200 - Success
  • 401 - Authentication failed
  • 404 - Calibre server not found / Book not found
  • 500 - Calibre server internal error
  • timeout - Request timeout (>30s)

CalibreAPIError Types

  • Connection failed - Network connectivity issue
  • Authentication failed - Invalid username/password
  • Server not found - Wrong URL or server not running
  • Request timeout - Server overloaded or slow response
  • JSON decode error - Malformed server response

Graceful Degradation

All MCP tools return valid responses even on errors:

  • Empty result lists for search failures
  • Error dictionaries with success: false and actionable error messages
  • Detailed suggestions for fixing issues

Webapp HTTP API

The Calibre webapp backend exposes REST endpoints (reservoir port 10720). See webapp/README.md.

LLM / AI Endpoints

Endpoint Method Description
/api/llm/models GET List models (Ollama/LM Studio/OpenAI). Query: provider, base_url
/api/llm/chat POST Chat completion. Body: messages, model, stream, provider, base_url

Config (backend .env): LLM_PROVIDER, LLM_BASE_URL, LLM_API_KEY

Core Endpoints

  • /api/books/ - List, add books
  • /api/authors/, /api/series/, /api/tags/ - List with pagination
  • /api/libraries/list, /api/libraries/switch, /api/libraries/stats
  • /api/search/ - Search books
  • /api/viewer/open-file - Open book in system default app
  • /api/export/csv, /api/export/json - Export books

🔧 Development Tools

MCP Inspector Testing

# Start MCP Inspector for interactive testing
python -m calibre_mcp.server
# Navigate to: http://127.0.0.1:6274

Direct API Testing

from calibre_mcp.calibre_api import quick_library_test
import asyncio

# Quick connection test
result = asyncio.run(quick_library_test("http://localhost:8080"))
print(f"Connection successful: {result}")

Configuration Testing

from calibre_mcp.config import CalibreConfig

config = CalibreConfig.load_config()
print(f"Server: {config.server_url}")
print(f"Auth configured: {config.has_auth}")

Built with Austrian efficiency for comprehensive e-book library management. Uses FastMCP 2.13+ portmanteau tool pattern for optimal Claude Desktop integration.