Credit to @mastrojs/atproto for the inspiration, in both the macro and the micro — check it out if you're more TypeScript inclined. Credit to @eaton for the name.

ATapult is a script that generates (or updates) Standard.Site records for you, allowing you to syndicate your website across the ATmosphere.

wilto@blues: ~/Sites/w-new on main $ npm run --silent atproto .well-known file found at ./src/.well-known/site.standard.publication/posts/index.html Publication 223nhvhpe22k2 https://wil.to/posts already exists. Documents 223numciw22k2 /automating-standard-site/ already exists. 223nu7gjd22k2 /standard-site/ already exists. 223nt3exb22k2 /llemdashes/ already exists. 223nsyskk22k2 /megamanathon-2/ already exists. 223nrxdf722k2 /googles-prompt-api/ already exists. 223nh6c4v22k2 /interesting-times/ already exists. 223my7e7722k2 /the-cloffice/ already exists. 223mdwi2322k2 /switching-to-a-framework-and-ubuntu/ already exists. Published successfully.

Be sure to check out Implementing Standard.Site for the general idea, use cases, and manual record-creation process, just to get your bearings — keep PDSIs and the Standard.Site Validator handy to validate your results.

Provide it with your PDS credentials, a Publication record, and your Document records, like so:

await ATapult({
 credentials: {
 identifier: "wil.to",
 password: process.env.ATPROTO_PASSWORD
 },
 url: new URL( "https://bsky.social" )
}, publicationRecord, documentRecords );

Have a look at the example to see how I'm using it on my website, which is: grab all the Markdown files from a directory, parse out their frontmatter, filter out any marked as a draft or a post that just links to an external article, and publish ’em.

First, you need credentials for either the Personal Data Server (PDS) provided by Bluesky (you can look yourself up on https://pdsls.dev) or your own. If you're using Bluesky's server, you can generate an app password at https://bsky.app/settings/app-passwords. Add said password to ATPROTO_PASSWORD=[xxxx-xxxx-xxxx-xxxx] in your .env file, which never leaves your local development environment in any way, right?

On the first run, this script asks you to confirm that your Publication and Document paths are as-expected — if so, it generates a .well-known file that corresponds with your specified Publication URL and the terminal output includes the link tags you'll need in the head of the Publication/Document pages (each page with the corresponding document rkey, of course). On subsequent runs, it creates new Document records with rkeys based on the dates you provide in each document's publishedAt field, and a Publication record with either an explicit rkey or one based on a publishedAt value. The terminal output includes the paths and record keys for your Publication and its Documents and whether they were created, updated with new information, or skipped because they exist and are unchanged.