The Testing area sorts 178 providers into API testing, mocking, load testing, contract testing and QA. 30 of them carry a score on the area page, and that scored group does well: 3 exemplar, 16 strong, 7 developing, 3 emerging and 1 thin, with a median of 56.0. Look at what the vendors themselves publish, though, and the record gets uneven. 14 of the 30 expose exactly one API.
The scored field
| Provider | Kin Score | Band | APIs | Agent Readiness |
|---|---|---|---|---|
| Postman | 78.0 | exemplar | 22 | 54.9 |
| Checkly | 72.5 | exemplar | 2 | 60.0 |
| Tricentis | 68.6 | exemplar | 9 | 44.6 |
| Speakeasy | 66.3 | strong | 2 | 53.3 |
| BrowserStack | 66.3 | strong | 1 | 44.1 |
| SmartBear | 55.9 | strong | 1 | 23.1 |
| Insomnia | 52.4 | developing | 1 | 19.5 |
| Bruno | 50.4 | developing | 3 | 9.6 |
| REST Assured | 27.6 | thin | 1 | 1.9 |
What the area tells you
Selling quality is not the same as publishing it. SmartBear stewarded Swagger and sells contract testing, and it scores 55.9 with one API and an Agent Readiness of 23.1, agent-aware. Insomnia and Bruno, two clients developers use to call other people’s APIs, publish little of their own: 52.4 and 50.4, with agent readiness under 20. These tools are built to inspect contracts, and they mostly do not hand out their own.
The ones that run in production score highest. Checkly runs synthetic monitoring, so its API is how customers actually use the product, and it has the area’s best Agent Readiness at 60.0. Postman is the only vendor here with a broad surface (22 APIs) to go with its score. The tools that live inside a developer’s editor rank lower than the ones that run on a schedule against production.
Libraries score like libraries. REST Assured (27.6), SuperTest (22.9) and Step CI (17.5) are open-source frameworks, not hosted services. A low Kin Score here does not mean they are poor tools. It means a Java DSL has no endpoint to score. Read those rows as category, not as a verdict.
One catalog note: the area claims 178 member providers, and its page renders score rows for only 30. The other 148 are real provider files. They just do not appear in the area’s standings yet. That gap is ours to close.
Takeaway
What would move this area is the testing vendors applying their own discipline to themselves: a published contract for each product, versioning signals, and an MCP or agent card so an agent can run a test suite rather than only read about one. Postman and Checkly already show it can be done.
Explore the area at testing.apievangelist.com.