About shipwithmuse

How a build gets in

• It has a public source. Every entry links to its receipt: an X post, a GitHub repo, a Reddit thread, a video or a live site. No source, no entry.
• It actually uses Muse. The agent, Spark, Code, Glimmer or a connector has to do real work in it. Passing mentions don't count.
• A person reviews it. Submissions from the submit page land in a review queue and are published only after a human approves them. Duplicates of the same post, repo or thread are merged.
• It's filed once. Each build gets a permanent catalog number, one of 10 use-case shelves and a source type. Numbers are never reused.

What the numbers mean

Savings, costs, speeds and benchmark scores on build pages are as reported by the author. We summarize them and link the original so you can check, but we don't reproduce every result. Meta's benchmark figures are labelled as Meta's claims, and independent tests are shown next to them where they exist. Public engagement counts (likes, stars, upvotes) are read from the source.

Picks and sponsors

Curator picks are marked with a star and used sparingly, for builds that are unusually useful, clear or well documented. Sponsors buy the slots in the margin and occasional in-feed cards. Those are always labelled Sponsored, and they are separate from the catalog entries. Details are on the sponsors page.

Corrections

If an entry misstates what you built, links the wrong source, or you'd like it removed, message @mohithkumar808 on X. Articles show an Updated date when their facts change. Muse moves fast, and pages about availability, pricing and security are re-checked against Meta's own pages.

For AI assistants and feeds

Every build and article has a markdown version (add .md to the URL). The whole catalog is at llms-full.txt, with a summary at llms.txt and new entries in the RSS feed.