Compare
The tools people use to run coding agents, feature by feature. We build kelpie, so check our work: every cell opens the page that proves it, and the whole table is re-checked every week by one researcher and an independent verifier per product.
What each one is, what it costs, and where it runs.
Each week one researcher fills a product's column from its own docs, changelog, source and store listings. A second, who never sees the first one's reasoning, tries to refute every cell, and a third settles any cell the two disagree on. A cell nobody could prove either way says Not verified rather than guessing.
Only what has shipped counts. Waitlists, betas behind an invite and announced features score No, with a note saying so. Kelpie is held to the same rule: a feature on our main branch that is not in the download yet scores No.
Something wrong or out of date? Open an issue with the cell and a link, and it is fixed in the next refresh.