將 Karakeep 連接至 Open Flow
Keep useful reading with the notes that make it easy to find again.
每月 20,000 次免費應用程式呼叫(付費服務另計)
應用程式介紹和工具詳情以英文提供。
可以做什麼
Search saved links and notes, read available article content, and organize bookmarks with tags. Existing bookmarks can be checked before saving the same link again.
Library Search and Reading
Find bookmarks and read their saved text, metadata, and highlights.
Bookmark Organization
Save links and notes, edit bookmark details, attach or remove tags, and maintain lists.
所有工具
Get Current User
Get the profile of the Karakeep user that owns the connected API key, including name, email and avatar.
karakeep.get_current_userGet Current User Stats
Get usage statistics for the Karakeep user that owns the connected API key, including bookmark, tag, list and asset counts.
karakeep.get_current_user_statsList Bookmarks
List the bookmarks of the connected Karakeep user, optionally filtered by archived or favourited status, sorted by creation date and paged with a cursor. Keep includeContent false unless the extracted page content is really needed, because it can make the response very large; text bookmarks always carry content.text.
karakeep.list_bookmarksCreate Bookmark
Create a bookmark from a link, a text note or an already uploaded asset. Set type to link and provide url, set type to text and provide text, or set type to asset and provide assetType and assetId. When the same URL is already bookmarked Karakeep does not create a duplicate: it returns the existing bookmark with alreadyExists set to true and re-saves it, which bumps createdAt to now, resets archived to false, and overwrites title, favourited, note and summary with whatever you supplied. Only source rss and source import skip that re-save. Call check_bookmark_url first when the existing bookmark must not be touched.
karakeep.create_bookmarkSearch Bookmarks
Search the bookmarks of the connected Karakeep user with full text, semantic or hybrid search. Paging uses a search specific cursor that cannot be shared with list_bookmarks. Keep includeContent false unless the extracted page content is really needed, because it can make the response very large; text bookmarks always carry content.text.
karakeep.search_bookmarksCheck Bookmark Url
Check whether a URL is already bookmarked and return the id of the existing bookmark, or null when it is not. Useful before calling create_bookmark.
karakeep.check_bookmark_urlGet Bookmark
Get a single bookmark with its tags, content and attached assets. Keep includeContent false unless the extracted page content is really needed, because it can make the response very large; text bookmarks always carry content.text. Use get_bookmark_content to read long article text in chunks.
karakeep.get_bookmarkUpdate Bookmark
Update a bookmark. Only the fields you send are changed, and sending null clears a nullable field, so existing values are overwritten. Fields that do not belong to the bookmark variant are not ignored: Karakeep rejects the whole request with a 400 such as Attempting to set link attributes for non-link type bookmark and writes nothing, so send only the fields that match the bookmark type.
karakeep.update_bookmarkDelete Bookmark
Permanently delete a bookmark together with its tags, highlights and attached assets.
karakeep.delete_bookmarkGet Bookmark Content
Read the readable content of a bookmark in chunks as markdown or plain text. This is the safe way to read long articles: pass nextCursor back to fetch the following chunk instead of asking for the whole content with includeContent. Karakeep rejects a cursor with CONTENT_CHANGED when the bookmark content changed while paging, in which case restart from the first chunk.
karakeep.get_bookmark_contentSummarize Bookmark
Generate an AI summary for a link bookmark and return it. Karakeep saves the generated summary on the bookmark, replacing any summary it already had, and re-indexes the bookmark for search, so this is not a read-only preview. The call blocks while the configured model runs, so it can take a while, and it fails with an invalid input error when the Karakeep instance has no inference provider configured. The current server writes the summary before answering and returns it synchronously, so summary normally carries the finished text. The published OpenAPI spec instead documents the whole bookmark record, so an instance following that shape can answer with summary null; in that case read the bookmark again with get_bookmark until summarizationStatus is success and take the summary from there.
karakeep.summarize_bookmarkAttach Tags To Bookmark
Attach tags to a bookmark. Reference each tag by tagId, or by tagName to let Karakeep create the tag when it does not exist yet.
karakeep.attach_tags_to_bookmarkDetach Tags From Bookmark
Detach tags from a bookmark. Reference each tag by tagId or by tagName. The tags themselves are kept, only the attachment to this bookmark is removed.
karakeep.detach_tags_from_bookmarkGet Bookmark Lists
List the Karakeep lists that contain a bookmark.
karakeep.get_bookmark_listsGet Bookmark Highlights
List the highlights stored on a bookmark.
karakeep.get_bookmark_highlightsAttach Asset To Bookmark
Attach an already uploaded asset to a bookmark, for example a screenshot or a PDF. Upload the file first to obtain an asset id. Only the screenshot, pdf, assetScreenshot, precrawledArchive, bannerImage, video and userUploaded roles can be attached; linkHtmlContent, bookmarkAsset, fullPageArchive, avatar and unknown assets are maintained by Karakeep itself and are rejected with a 400.
karakeep.attach_asset_to_bookmarkReplace Asset On Bookmark
Replace an asset that is attached to a bookmark with another already uploaded asset. The replaced asset is deleted and the new asset takes over its role. The asset being replaced must hold an attachable role, meaning screenshot, pdf, assetScreenshot, precrawledArchive, bannerImage, video or userUploaded; assets Karakeep maintains itself, including linkHtmlContent, bookmarkAsset, fullPageArchive, avatar and unknown, are rejected with a 400.
karakeep.replace_asset_on_bookmarkDetach Asset From Bookmark
Detach an asset from a bookmark. Karakeep deletes the asset record together with the stored file, so the asset id becomes unusable afterwards and the bytes are gone; the bookmark keeps its other assets. Only screenshot, pdf, assetScreenshot, fullPageArchive, precrawledArchive, bannerImage, video and userUploaded assets can be detached; linkHtmlContent, bookmarkAsset, avatar and unknown assets are maintained by Karakeep itself and are rejected with a 400.
karakeep.detach_asset_from_bookmarkList Lists
Retrieve all bookmark lists for the authenticated Karakeep user, including both manual and smart lists. This endpoint is not paginated and returns every list in one response.
karakeep.list_listsCreate List
Create a new bookmark list. Manual lists receive bookmarks that are added explicitly, while smart lists are populated automatically by a saved search query.
karakeep.create_listGet List
Retrieve a single Karakeep list by its id.
karakeep.get_listUpdate List
Partially update a Karakeep list. Only the fields present in the input are changed, and sending null for description or parentId clears the stored value.
karakeep.update_listDelete List
Delete a Karakeep list. Only the list itself is removed; the bookmarks it contained are kept.
karakeep.delete_listGet List Bookmarks
Retrieve one page of the bookmarks inside a Karakeep list. For smart lists the bookmarks are computed from the saved query of the list.
karakeep.get_list_bookmarksAdd Bookmark To List
Add a bookmark to a manual Karakeep list. The operation is idempotent, so adding a bookmark that is already in the list succeeds and changes nothing.
karakeep.add_bookmark_to_listRemove Bookmark From List
Remove a bookmark from a manual Karakeep list. The bookmark itself is kept. Karakeep rejects the request when the bookmark is not a member of the list.
karakeep.remove_bookmark_from_listList Tags
Retrieve one page of Karakeep tags. Tags can be filtered by name fragment and by who attached them, and sorted by name, usage count or relevance. Pagination is only active when limit is given: omit limit and Karakeep returns every tag in one response with no cursor.
karakeep.list_tagsCreate Tag
Create a new Karakeep tag. The name is trimmed and normalized into the tag style configured for the account.
karakeep.create_tagGet Tag
Retrieve a single Karakeep tag by its id, including how many bookmarks carry it and who attached it.
karakeep.get_tagUpdate Tag
Rename a Karakeep tag. The name is the only field this endpoint can change, so it is required. The new name is trimmed and normalized, and every bookmark carrying the tag sees the new name.
karakeep.update_tagDelete Tag
Delete a Karakeep tag. The tag is detached from every bookmark that carried it; the bookmarks themselves are kept.
karakeep.delete_tagGet Tag Bookmarks
Retrieve one page of the bookmarks that carry a given Karakeep tag.
karakeep.get_tag_bookmarksList Highlights
Retrieve one page of the highlights the authenticated Karakeep user has made across all bookmarks.
karakeep.list_highlightsCreate Highlight
Create a text highlight on a Karakeep bookmark. A highlight is defined by the character offsets it covers in the readable content of the bookmark and can carry a color and a note.
karakeep.create_highlightGet Highlight
Retrieve a single Karakeep highlight by its id.
karakeep.get_highlightUpdate Highlight
Partially update a Karakeep highlight. Only the color and the note can be changed, and sending null for the note clears it.
karakeep.update_highlightDelete Highlight
Delete a Karakeep highlight and return the record that was removed. The bookmark the highlight belonged to is kept.
karakeep.delete_highlightUpload Asset
Upload a file to Karakeep as a new standalone asset. The connector downloads fileUrl server side and posts the bytes as multipart/form-data under the form field file. Karakeep accepts only image/gif, image/jpeg, image/png, image/webp, video/mp4, video/webm, video/x-matroska, text/html and application/pdf. It sniffs binary types from the bytes themselves and ignores the declared type for them, but HTML carries no signature, so an HTML upload is accepted only when the declared content type is exactly text/html. Anything above the instance asset size limit is rejected with an Asset is too big error; that limit is MAX_ASSET_SIZE_MB multiplied by 1024 * 1024, so it is 50 MiB by default. The asset is created detached from any bookmark, so follow up with attach_asset_to_bookmark when it belongs to one.
karakeep.upload_assetGet Asset
Download the binary content of a Karakeep asset and hand it back as a temporary connector file URL. The connector streams the bytes from Karakeep into file transit instead of returning them inline, so the action also works for large images, videos, PDFs and archives. Karakeep sends no Content-Disposition header, so the file name falls back to the asset id plus an extension guessed from the content type.
karakeep.get_assetGet Asset Signed Url
Create a short lived signed download URL for a Karakeep asset. The URL carries its own token, so anyone holding it can fetch the asset without an API key, and it expires between 15 and 75 minutes after it was issued. Karakeep builds the URL from the public URL configured on the instance itself rather than from the instance URL stored on this connection, so a misconfigured self-hosted instance can return a host that is unreachable from the outside.
karakeep.get_asset_signed_urlList Backups
List every account backup recorded for the connected Karakeep user, including the ones that are still pending and the ones that failed.
karakeep.list_backupsCreate Backup
Trigger a new full account backup for the connected Karakeep user. Karakeep records the request and answers immediately with a backup whose status is pending; the archive itself is produced asynchronously by the instance backup worker, and assetId, size and bookmarkCount stay empty until it finishes. Poll get_backup until status becomes success or failure. When the backup worker is disabled on that instance the record stays pending forever, so always give the polling loop a timeout of its own. Karakeep rate limits this to five backups per hour.
karakeep.create_backupGet Backup
Get one Karakeep backup by id, including its current status, archive size, bookmark count and failure message. This is the polling target for create_backup: keep reading it until status is success or failure, and treat a record that stays pending as a backup worker that is not running on that instance.
karakeep.get_backupDelete Backup
Permanently delete a Karakeep backup record together with the archive file it produced. Karakeep answers with an empty body, so the action reports the deleted backup id instead.
karakeep.delete_backupDownload Backup
Download the archive of a finished Karakeep backup and hand it back as a temporary connector file URL. The action reads the backup record first: while the status is still pending it fails with an invalid input error asking you to poll get_backup, on a failed backup it fails with an invalid input error carrying the upstream failure message because that state is terminal and polling will not help, and a success record without an archive asset fails as a provider error. Only then does it stream the archive out of Karakeep into file transit. Karakeep's own download endpoint answers with a redirect to an authenticated relative path that most HTTP clients cannot follow, so the connector resolves the archive asset id itself instead of following the redirect.
karakeep.download_backupList Feeds
Retrieve all RSS feed subscriptions for the authenticated Karakeep user. The response is not paginated and returns every feed at once.
karakeep.list_feedsCreate Feed
Create a new RSS feed subscription. Karakeep periodically fetches the feed and imports matching items as bookmarks. Karakeep answers with 400 when the per user feed limit has already been reached.
karakeep.create_feedGet Feed
Retrieve a single RSS feed subscription by its id.
karakeep.get_feedUpdate Feed
Update an RSS feed subscription. Only the fields present in the input are changed; every omitted field keeps its current value.
karakeep.update_feedDelete Feed
Delete an RSS feed subscription. Bookmarks that the feed already imported are not affected.
karakeep.delete_feedFetch Feed Now
Trigger an immediate fetch of an RSS feed subscription. The fetch is only enqueued and runs asynchronously, so newly imported bookmarks appear later; poll get_feed and watch lastFetchedAt to see when it finished.
karakeep.fetch_feed_nowAdmin Update User
Update another Karakeep user's role, bookmark quota, storage quota or browser crawling setting. Only the provided fields are changed, and at least one of them must be present. Karakeep rejects updating the user that owns the API key with a 400. Requires an API key owned by a Karakeep admin user; any other key fails with a 403 provider_error carrying the upstream message Forbidden - Admin access required.
karakeep.admin_update_userAdmin Trigger Recrawl
Trigger a recrawl of link bookmarks across the whole Karakeep instance. Filter by crawl status to target specific bookmarks, for example only the failed ones, and by how recently they were modified. The crawls are queued and run asynchronously. Requires an API key owned by a Karakeep admin user; any other key fails with a 403 provider_error carrying the upstream message Forbidden - Admin access required.
karakeep.admin_trigger_recrawlAdmin Trigger Reindex
Trigger a reindex of bookmarks in the Karakeep search engine. Without modifiedWithinSeconds Karakeep clears the whole search index first and then re-queues every bookmark, so search results across the instance stay incomplete until the queue drains; with it only the bookmarks modified inside that window are re-queued and the existing index is preserved. The reindex runs asynchronously. Requires an API key owned by a Karakeep admin user; any other key fails with a 403 provider_error carrying the upstream message Forbidden - Admin access required.
karakeep.admin_trigger_reindexAdmin Trigger Inference
Trigger AI inference, either tagging or summarization, on bookmarks across the whole Karakeep instance. Filter by inference status and by how recently the bookmarks were modified. The inference jobs are queued and run asynchronously. Requires an API key owned by a Karakeep admin user; any other key fails with a 403 provider_error carrying the upstream message Forbidden - Admin access required.
karakeep.admin_trigger_inference沒有符合搜尋條件的工具。
將 Karakeep 連接至 Open Flow
每月 20,000 次免費應用程式呼叫(付費服務另計)