Import your X bookmarks into Readwise

Readwise has had its own path for capturing posts from X for years. If you have it enabled, importing this export on top of it gives you the same posts twice, attributed to two different sources, and spaced repetition will happily show you both. Worth deciding which one is the source of truth before you upload anything.

They capture different things, which is the argument for choosing rather than merging. Readwise’s own integration gets the posts you deliberately send it. A Sift export is your bookmark list — everything you tapped save on, including the several hundred you have no memory of — with a category, a set of tags and a one-line summary attached to each. If your bookmarks are a to-read pile rather than a curated collection, that difference is the reason to import them here at all: the categories are what make an unsorted pile reviewable.

1. 1

   Export from Sift and open the file

   You get bookmarks-YYYY-MM-DD.csv, newest saved first. The full column reference covers what each field holds; the only thing that matters here is that the headers are ours, and Readwise keys on its own.
2. 2

   Rename the headers to Readwise's

   Readwise’s CSV importer matches on column names. At minimum it needs a highlight column; it also understands a title, an author, a source URL, a note and a date. Their import page lists the current set and it is theirs to change, so read it before you upload rather than after. The export converter will rename and drop columns for you without the file leaving your browser.
3. 3

   Decide what a source is

   Everything in one upload lands as one source. Splitting the file by ai_category and uploading each part with its own title — X bookmarks: Engineering, X bookmarks: Business — costs a few minutes and makes the library navigable afterwards. Twelve sources of forty highlights beat one source of five hundred.
4. 4

   Prefix the note field

   Put ai_summary in the note, but write Sift summary: … rather than the bare sentence. Readwise notes render directly beneath the highlight, and an unlabelled generated line sitting under a quoted post reads as part of it.
5. 5

   Upload, then check one highlight

   Open a highlight you recognise and confirm the source URL resolves to the right post and the text was not truncated at the first line break. Posts with hard newlines are where CSV imports most often go wrong.

Review will start showing you posts. Readwise’s daily review draws from every source, so a few hundred saved posts change the character of it noticeably. If you want the archive without the drip, exclude the source from review — it is a per-source setting, and much easier than deciding later that the import was a mistake.

A copy in Readwise is outside our reach. When a post is deleted on X we are obliged to remove it here, and a nightly deletion check is how that happens — which also means the export never contains a post that has already gone. But a highlight you uploaded last month is in your Readwise account, not ours; nothing we do reaches it. That is not a warning, just the honest shape of exporting anything to anywhere.

Readwise has a documented API that accepts highlights, and this is the destination where a direct integration would be most obviously useful. We have not built one.

The X developer policy restricts providing X content to third parties, permitting the redistribution of post IDs and granting a narrow exception for up to 500 post objects per person per day by non-automated means. A scheduled server-to-server push of post text into another product is not that exception, and reading it as though it were would be optimistic. If we ever did build it, the compliant shape would send post IDs plus our categories and summaries and leave the text behind — which for a highlights app is the one field that matters. Until that changes, the file is the boundary; the bookmarks endpoint reference covers what we read and under whose token.