komga-mcp
MCP server exposing Komga's REST API as tools for browsing and managing comic, manga, BD, magazine, and ebook libraries.
README
komga-mcp
Part of the arr-mcps collection. MCP server exposing Komga's REST API as tools for browsing and managing comic, manga, BD, magazine, and ebook libraries. It is built with FastMCP.
Install
Download a wheel from the latest release
and install it as a uv tool:
uv tool install komga_mcp-*.whl
Register it with Claude Code:
claude mcp add komga \
--env KOMGA_URL=https://your-komga-host \
--env KOMGA_API_KEY=<api-key> \
-- komga-mcp
From source
uv sync
cp .env.example .env
claude mcp add komga \
--env KOMGA_URL=https://your-komga-host \
--env KOMGA_API_KEY=<api-key> \
-- uv run --directory /path/to/komga-mcp komga-mcp
Configuration
Komga API keys are created in the Komga user settings. The key is sent as
X-API-Key on every request. The server does not accept credentials supplied
by an MCP tool call.
| Env var | Required | Default |
|---|---|---|
KOMGA_URL |
yes | - |
KOMGA_API_KEY |
yes | - |
KOMGA_VERIFY_TLS |
no | true |
KOMGA_URL is the server root, without /api; the client adds the documented
/api/v1 and /api/v2 paths itself. Set KOMGA_VERIFY_TLS=false only when a
self-signed certificate is intentional.
Tools
8 resource-scoped tools, each covering multiple Komga REST endpoints
(81 total) via an operation parameter. Call a tool with operation set
to one of its listed operations (the function names below, e.g.
list_libraries) and an arguments dict matching that operation's
parameters — the tool's own description (visible to your MCP client)
lists every operation, its signature, and a one-line doc, including any
"Requires ADMIN" notes. Image, file-download, page-stream, and multipart
ComicRack endpoints are intentionally excluded.
| Tool | Operations | Kind | Covers |
|---|---|---|---|
komga_libraries |
9 | reads + writes | Libraries: list, get, create, update, delete, scan, analyze, refresh metadata, empty trash |
komga_series |
13 | reads + writes | Series: search, get, collections, latest/new/updated, thumbnails, metadata, analyze, read progress |
komga_books |
16 | reads + writes | Books: search, get, pages, next/previous, readlists, thumbnails, latest/ondeck/duplicates, metadata, analyze, read progress |
komga_collections |
6 | reads + writes | Collections: list, get, series, create, update, delete |
komga_readlists |
8 | reads + writes | Readlists: list, get, books, next/previous book, create, update, delete |
komga_referential_metadata |
10 | read-only | Authors, tags, genres, publishers, languages, age ratings, sharing labels, release years |
komga_users_api_keys |
12 | reads + writes | whoami, users CRUD, passwords, API keys, auth activity |
komga_server |
7 | reads + writes | Server info, settings, claim status, history, tasks, filesystem |
Example: komga_libraries(operation="scan_library", arguments={"library_id": "l1"}).
Search conditions
search_series, search_books, and search_series_alphabetical_groups send a
JSON body containing fullTextSearch and an optional recursive condition.
Pagination is sent as query parameters (page, size, repeated sort, and
unpaged). For example:
{
"fullTextSearch": "berserk",
"condition": {
"allOf": [
{"genre": {"operator": "contains", "value": "action"}},
{"readStatus": {"operator": "is", "value": "UNREAD"}}
]
}
}
Use anyOf for OR and allOf for AND. String operators include is, isNot,
beginsWith, contains, endsWith, and their negative forms. Boolean fields
use isTrue or isFalse; date fields use after, before, isInTheLast,
isNotInTheLast, isNull, or isNotNull.
Metadata patches use Komga's normal semantics: omit a field to leave it alone,
or send null to unset it. Read progress accepts page and/or completed.
Development
make help
| Command | Does |
|---|---|
make sync |
Run uv sync |
make test |
Run the offline test suite |
make test-integration |
Run live tests (needs KOMGA_URL/KOMGA_API_KEY) |
make build |
Build wheel and sdist into dist/ |
make bump-patch / bump-minor / bump-major |
Bump package version and lockfile |
make clean |
Remove build artifacts |
The release workflow runs tests and publishes wheel and sdist files whenever a
v* tag is pushed.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。