Automatically bookmark X posts about the topics you follow

Describe what deserves a second read

Start with one topic: “Original posts explaining open-source machine learning research with concrete findings or benchmark results.” Add an exception for affiliate links, promotional tool lists and unsupported hype.

A narrow rule is easier to review than “save anything interesting.” Choose the AI research template in the rule library, then adapt it to your own field.

Select Bookmark, not Like

In Jev, select the Bookmark action. A matching post is saved to X Bookmarks. Like is a separate action and does not create a bookmark.

The action uses your X session. It does not create a separate ReDeck reading list or bookmark folder.

Preview recent saved posts

Connect your TypeSafe API key and enable Timeline History. Enter the rule and exception, choose a starting threshold, then preview recent saved posts.

Inspect examples that should match and examples that should not. The model only receives post text, author handle and display name; it does not read the full article behind a link. A short post may therefore omit the detail your rule needs.

Enable and review the queue

After saving and enabling the rule, review Jev’s activity log for completed bookmarks and errors. New or edited rules do not automatically act on your entire old history.

Request and per-minute action limits can delay processing. A matching score is not the same as a successful bookmark; check the recorded action status.

The website’s rule library and builder only prepare settings. Actual evaluation uses TypeSafe API requests, including when you preview a rule.