Karakeep MCP server
The Karakeep MCP server lets you manage bookmarks, lists, and tags in Karakeep.
Bookmarks
Search (
search-bookmarks) — Full-text search with qualifiers (favorites, archived, tags, lists, date ranges, URL substrings, types, etc.)Get a bookmark (
get-bookmark) — Retrieve a bookmark by IDGet bookmark content (
get-bookmark-content) — Retrieve the full content of a bookmarkCreate a text bookmark (
create-text-bookmark) — Save text as a bookmark with an optional titleCreate a URL bookmark (
create-url-bookmark) — Save a URL as a bookmark with an optional titleUpdate or delete bookmarks — Modify or remove existing bookmarks
Lists
Get all lists (
get-lists) — Retrieve all available listsGet a single list — Retrieve a specific list by ID
Get bookmarks in a list (
get-bookmarks-in-list) — Retrieve bookmarks belonging to a listCreate a list (
create-list) — Create a list with a name, emoji icon, optional parent, description, query, and public flagUpdate or delete a list — Modify or remove a list
Add bookmark to a list (
add-bookmark-to-list) — Add a bookmark to a specific listRemove bookmark from a list (
remove-bookmark-from-list) — Remove a bookmark from a list
Tags
List tags — Retrieve tags with filters, pagination, and usage counts
Get a single tag — Retrieve a specific tag by ID
Rename or delete tags — Modify or remove existing tags
List bookmarks by tag — Retrieve bookmarks attached to a specific tag
Attach tags (
attach-tag-to-bookmark) — Attach one or more tags to a bookmarkDetach tags (
detach-tag-from-bookmark) — Remove one or more tags from a bookmark
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Karakeep MCP server search for bookmarks about machine learning"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Karakeep MCP Server
This is the Karakeep MCP server, which is a server that can be used to interact with Karakeep from other tools.
Supported Tools
Bookmarks
Searching bookmarks (
search-bookmarks)Reading a bookmark (
get-bookmark,get-bookmark-content)Listing the lists that contain a bookmark (
get-bookmark-lists)Creating text and URL bookmarks (
create-bookmark)Updating a bookmark (
update-bookmark)Deleting a bookmark (
delete-bookmark)
Lists
Listing all lists (
get-lists)Retrieving a single list (
get-list)Creating a list (
create-list)Updating a list (
update-list) — name, icon, description, parentId, query, public; field constraints (length caps, smart-query validation) are inherited from the shared schemaDeleting a list (
delete-list) — child lists are NOT deleted with the parent; they become root-level listsListing bookmarks in manual and smart lists (
get-list-bookmarks)Adding and removing bookmarks from lists (
add-bookmark-to-list,remove-bookmark-from-list)
Tags
Listing tags with filters / pagination (
get-tags)Retrieving a single tag with usage counts (
get-tag)Renaming a tag (
update-tag)Deleting a tag (
delete-tag) — bookmarks that had the tag are not deletedListing the bookmarks attached to a tag (
get-tag-bookmarks)Attaching and detaching tags on bookmarks (
attach-tag-to-bookmark,detach-tag-from-bookmark)
Assets
Getting a temporary signed download URL for an asset (
get-asset)
Highlights
Listing highlights across all bookmarks (
list-highlights)Listing highlights for a bookmark (
get-bookmark-highlights)Retrieving a single highlight (
get-highlight)Creating, updating, and deleting highlights (
create-highlight,update-highlight,delete-highlight)
Currently, the MCP server only exposes tools (no resources).
Related MCP server: Bookmark Geni MCP Server
Usage with Claude Desktop
From NPM:
{
"mcpServers": {
"karakeep": {
"command": "npx",
"args": [
"@karakeep/mcp"
],
"env": {
"KARAKEEP_API_ADDR": "https://<YOUR_SERVER_ADDR>",
"KARAKEEP_API_KEY": "<YOUR_TOKEN>",
"KARAKEEP_CUSTOM_HEADERS": "{\"CF-Access-Client-Id\": \"...\", \"CF-Access-Client-Secret\": \"...\"}"
}
}
}
}From Docker:
{
"mcpServers": {
"karakeep": {
"command": "docker",
"args": [
"run",
"-e",
"KARAKEEP_API_ADDR=https://<YOUR_SERVER_ADDR>",
"-e",
"KARAKEEP_API_KEY=<YOUR_TOKEN>",
"-e",
"KARAKEEP_CUSTOM_HEADERS={\"CF-Access-Client-Id\": \"...\", \"CF-Access-Client-Secret\": \"...\"}",
"ghcr.io/karakeep-app/karakeep-mcp:latest"
]
}
}
}Available Tools
29 toolsadd-bookmark-to-listC
Add a bookmark to a list.
Inputs: listId (required), bookmarkId (required).
attach-tag-to-bookmarkC
Attach a tag to a bookmark.
Inputs: bookmarkId (required), tagsToAttach (required).
create-bookmarkC
Create a link bookmark or a text bookmark
Inputs: type (required), content (required), title.
create-highlightA
Create a text highlight on a bookmark using character offsets from the bookmark's readable content.
Inputs: note (required), text (required), endOffset (required), bookmarkId (required), +2 more.
create-listC
Create a list.
Inputs: icon (required), name (required), parentId.
delete-bookmarkA
Delete a bookmark by id. This is destructive — the bookmark, its highlights, and its assets are removed.
Inputs: bookmarkId (required).
delete-highlightCDestructive
Delete a highlight by id.
Inputs: highlightId (required).
delete-listA
Delete a list by id. Bookmarks inside the list are not deleted. Child lists are also not deleted. Any child lists become root-level lists (their parentId is set to null). If that isn't the tree change you want, move or re-parent the children before calling this.
Inputs: listId (required).
delete-tagB
Delete a tag by id. Bookmarks that had this tag are not deleted; the tag is just removed from them.
Inputs: tagId (required).
detach-tag-from-bookmarkC
Detach a tag from a bookmark.
Inputs: bookmarkId (required), tagsToDetach (required).
get-assetA
Get a temporary signed URL for downloading an asset by its asset ID.
Inputs: assetId (required).
get-bookmarkC
Get a bookmark by id.
Inputs: bookmarkId (required).
get-bookmark-contentB
Get the content of the bookmark in markdown
Inputs: bookmarkId (required).
get-bookmark-highlightsCRead-only
List every highlight on a bookmark.
Inputs: bookmarkId (required).
get-bookmark-listsBRead-only
List every list that contains a bookmark.
Inputs: bookmarkId (required).
get-highlightBRead-only
Retrieve a single highlight by id.
Inputs: highlightId (required).
get-listB
Retrieve a single list by its id.
Inputs: listId (required).
get-list-bookmarksARead-only
List bookmarks in a list by its stable id. Smart lists are evaluated using their saved query.
Inputs: listId (required), limit, cursor, sortOrder, +1 more.
get-listsB
Retrieves a list of lists.
Inputs: no named parameters.
get-tagA
Retrieve a single tag by id, including its bookmark counts.
Inputs: tagId (required).
get-tag-bookmarksB
List bookmarks that have a given tag, with pagination.
Inputs: tagId (required), limit, cursor, sortOrder, +1 more.
get-tagsB
List tags with their bookmark counts. Supports filtering and pagination.
Inputs: sort, limit, cursor, attachedBy, +1 more.
list-highlightsARead-only
List highlights across all bookmarks, newest first.
Inputs: limit, cursor.
remove-bookmark-from-listC
Remove a bookmark from a list.
Inputs: listId (required), bookmarkId (required).
search-bookmarksC
Search for bookmarks matching a specific query using full-text, semantic, or hybrid search.
Inputs: query (required), limit, sortOrder, nextCursor, +1 more.
update-bookmarkA
Update fields on an existing bookmark. Only the fields you pass are modified; omitted fields stay unchanged. Returns the updated bookmark.
Inputs: bookmarkId (required), url, note, title, +7 more.
update-highlightB
Update a highlight's color or note.
Inputs: highlightId (required), note, color.
update-listA
Update a list. Only the fields you pass are changed. Length caps and smart-list query rules come from the shared list schema.
Inputs: listId (required), icon, name, query, +3 more.
update-tagB
Rename a tag. The new name is normalized by the server.
Inputs: name (required), tagId (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 29 tools
Most tools map cleanly to a distinct resource+action (bookmark, list, tag, highlight, asset), and descriptions disambiguate the tricky pairs. However, the inverse-named pair get-bookmark-lists (lists containing a bookmark) vs get-list-bookmarks (bookmarks in a list) is easy to swap, and get-tags/get-tag plus list-highlights/get-bookmark-highlights sit close together.
Kebab-case verb_noun is used throughout, which is good, but the verb for collection retrieval is inconsistent: get-tags, get-lists, get-bookmark-lists use 'get' while list-highlights and get-list-bookmarks use 'list'/'get' in different orders. Noun ordering also flips between get-bookmark-lists and get-list-bookmarks, which undermines predictability in an otherwise uniform scheme.
At 29 tools the surface is on the heavy side, but the domain genuinely spans five resources (bookmarks, lists, tags, highlights, assets) with real CRUD, so most tools earn their place. A few could be collapsed (e.g. the collection-level getters) to trim the count.
Bookmarks, lists, tags, and highlights all have near-complete lifecycle coverage, including membership add/remove and search. The gaps are around assets: get-asset exists but nothing lists a bookmark's asset IDs or uploads assets, and there is no create-tag (tags only appear via attach), leaving minor dead ends agents must work around.
Maintenance
Related MCP Connectors
All your bookmarks in one library. Share them with your AI agents. Read-only search over your saves.
Search, read and save to your private vault of images, links and notes.
Save, search, read, tag, and organize your bookmarks from AI assistants with OAuth sign-in.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables interaction with Anybox bookmark manager on macOS, allowing users to search, save, and organize bookmarks with tags and folders through natural language.16-
- AlicenseNot gradedqualityDmaintenanceEnables semantic search across browser bookmarks from Chrome, Firefox, Edge, Opera, and other browsers using natural language queries. Extracts and indexes bookmark content and metadata into a vector database for intelligent retrieval.1Apache 2.0
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to save and search bookmarks using OpenAI's RAG capabilities for intelligent bookmark management and retrieval.MIT
- AlicenseBqualityDmaintenanceEnables AI assistants to save and search bookmarks with semantic search using OpenAI, allowing storage of URLs with metadata and intelligent retrieval across collections.235 PyPIMIT