APIs & Testing

The tooling around APIs has quietly become one of the healthier corners of developer software, mostly because the problems are well understood and the feedback is fast. The launches here cover clients for making requests, mocking and stubbing, contract and schema testing, documentation generation, and the growing set of tools that watch a live API and tell you when it changed.

For clients, the meaningful axis is not features — every client can send a request — it is how well it handles the boring parts of real work: environments and secrets, sharing a collection with a colleague without leaking a token, scripting a sequence of dependent calls, and diffing two responses. Tools that store collections as plain files in your repository are markedly easier to live with than ones that keep them in a proprietary cloud, because they version, review and merge like everything else you own.

Mocking is the most underused thing in this subcategory. Being able to develop a front end against a mock that matches the contract, before the back end exists, removes the single most common source of integration delay. The good mocking tools generate from a schema rather than from recorded traffic, so the mock stays honest as the contract evolves.

Contract testing is the grown-up version of the same idea and is where teams find bugs that unit tests structurally cannot see. The value is not in catching a broken endpoint — you will notice that — it is in catching the field that silently changed type, or the optional property that quietly became required. Tools that fail loudly on both directions of drift are doing the job.

Documentation generators are judged almost entirely on whether the output stays true. Docs generated from the schema at build time stay correct. Docs written by hand alongside the schema do not, no matter how good the intention, and the gap between them is where support tickets come from.

Pricing here is usually per-seat and generally sane, with the exception of anything hosting your mocks or monitors, where the meter is request volume. If you are running a monitor every thirty seconds against twenty endpoints, do the multiplication before you commit.