starred/karakeep

Fork 0

mirror of https://github.com/karakeep-app/karakeep.git synced 2026-04-25 16:06:04 +03:00

[GH-ISSUE #2041] AI-powered search that helps provide the right bookmark based on a description #1267

New issue

Open

opened 2026-03-02 11:56:10 +03:00 by kerem · 2 comments

kerem commented

2026-03-02 11:56:10 +03:00

Owner

Originally created by @bluekeybo on GitHub (Oct 13, 2025).
Original GitHub issue: https://github.com/karakeep-app/karakeep/issues/2041

Describe the feature you'd like

It'd be very useful to have an AI-powered search where the user can ask a detailed description of the bookmark they're looking for and it returns some possible bookmarks that meet that criteria. Maybe on the backend the tool searches through the links, tags, and the AI-summaries of each link.

Example use case:

Search input:
"Book about electronics, that's considered to be a fundamental reference for anyone interested in circuits or engineering"

Output:
"Here's what I found from your bookmarks:
https://en.wikipedia.org/wiki/The_Art_of_Electronics
https://en.wikibooks.org/wiki/Amateur_Radio_Manual"

Describe the benefits this would bring to existing Karakeep users

The main use-case for Karakeep is to archive/store/save bookmarks so you can easily find/refer to them later. Currently, the search tool is not very useful, and it doesn't find items unless you're very specific. An AI-powered search for those who have already linked AI in the backend would be a tremendous quality improvement for anyone using Karakeep as it would help users find bookmarks based on a description that may or may not contain the exact tag/link of the bookmark.

Can the goal of this request already be achieved via other means?

No, current search tool is underpowered.

Have you searched for an existing open/closed issue?

I have searched for existing issues and none cover my fundamental request

Additional context

No response

Originally created by @bluekeybo on GitHub (Oct 13, 2025). Original GitHub issue: https://github.com/karakeep-app/karakeep/issues/2041 ### Describe the feature you'd like It'd be very useful to have an AI-powered search where the user can ask a detailed description of the bookmark they're looking for and it returns some possible bookmarks that meet that criteria. Maybe on the backend the tool searches through the links, tags, and the AI-summaries of each link. Example use case: Search input: "Book about electronics, that's considered to be a fundamental reference for anyone interested in circuits or engineering" Output: "Here's what I found from your bookmarks: https://en.wikipedia.org/wiki/The_Art_of_Electronics https://en.wikibooks.org/wiki/Amateur_Radio_Manual" ### Describe the benefits this would bring to existing Karakeep users The main use-case for Karakeep is to archive/store/save bookmarks so you can easily find/refer to them later. Currently, the search tool is not very useful, and it doesn't find items unless you're very specific. An AI-powered search for those who have already linked AI in the backend would be a tremendous quality improvement for anyone using Karakeep as it would help users find bookmarks based on a description that may or may not contain the exact tag/link of the bookmark. ### Can the goal of this request already be achieved via other means? No, current search tool is underpowered. ### Have you searched for an existing open/closed issue? - [x] I have searched for existing issues and none cover my fundamental request ### Additional context _No response_

kerem added the

feature request

status/untriaged

labels

2026-03-02 11:56:10 +03:00

kerem commented

2026-03-02 11:56:11 +03:00

Author

Owner

@MohamedBassem commented on GitHub (Oct 13, 2025):

This is planned, but for some reason I can't find the original issue that suggested this :)

@MohamedBassem commented on GitHub (Oct 13, 2025): This is planned, but for some reason I can't find the original issue that suggested this :)

kerem commented

2026-03-02 11:56:11 +03:00

Author

Owner

@DeepanshKhurana commented on GitHub (Oct 13, 2025):

Interesting idea! Since I have been working on AI-inference applications, I have found that AI works best as an interpretor, at least in cases like these. My suggestion would be that instead of circumventing Meillisearch (since it does have its uses and I'm sure users will miss the keyword-based searching) to instead go from a prompt to Meillisearch syntax pipeline. So, the prompt takes the query, breaks it down into a fine-tuned Meillisearch query with AI's help and on the backend run this new parsed query. Maybe an interpretive layer of "here is how your search was interpreted" could be a great idea to go along with it.

@DeepanshKhurana commented on GitHub (Oct 13, 2025): Interesting idea! Since I have been working on AI-inference applications, I have found that AI works best as an interpretor, at least in cases like these. My suggestion would be that instead of circumventing Meillisearch (since it does have its uses and I'm sure users will miss the keyword-based searching) to instead go from a prompt to Meillisearch syntax pipeline. So, the prompt takes the query, breaks it down into a fine-tuned Meillisearch query with AI's help and on the backend run this new parsed query. Maybe an interpretive layer of "here is how your search was interpreted" could be a great idea to go along with it.

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

starred/karakeep#1267

No description provided.

Rows
Columns