You are three weeks into picking a pipeline tool for your team. You have a spreadsheet half filled in, a manager who wants a shortlist by Friday, and a habit of starting with a catalog page rather than a vendor's marketing site. So you open CI/CD & Deployment Automation on APIs.io. It lists 64 providers, and the top of the list looks right. GitHub Actions is first at 82.8. GitLab is second at 82.6.
Then you hit third place. It's Salesforce, at 82.3, ahead of Harness at 79.3. That's odd for someone shopping for continuous integration. You look for CircleCI, Jenkins and Buildkite. They aren't on the page at all. Neither is LaunchDarkly, even though all four are in the catalog.
The APIs.io team has published a post explaining what you're looking at, and it doesn't blame you or the vendors. It says this is a defect in how they build the page.
Here is the cause. Membership in a use-case page comes from each provider's free-text use_cases field, matched against aliases. It does not look at tags. The post counts 164 scored providers tagged CI/CD, continuous integration, continuous delivery or continuous deployment. Only 12 of them are on the page. Of the 152 left out, 145 have no use_cases entries at all, so the build has nothing to match. The other 52 members got in through a sentence in that field.
The left-out list reads like your shortlist. LaunchDarkly scores 80.2, with agent readiness of 54.1. Azure DevOps is 74.4. CircleCI is 72.6, and its agent score is higher than GitLab's, yet it's missing while GitLab sits second. Buildkite is 71.8, CloudBees 71.4, Octopus Deploy 67.2, Akuity 66.0 and Jenkins 40.7. The post notes that LaunchDarkly would place fourth, above Harness, and that the page's own definition even names feature flags.
Then there's who got in instead. Salesforce earned its place with one entry, "DevOps And CI/CD Automation," about its metadata deployment APIs. The post calls that true, but not what someone looking for CI wants. Palo Alto Networks arrives through "SASE Deployment Automation" and "DevSecOps Pipeline Integration," and brings 468 APIs, which is 53% of the 883 on the page. Autodesk Maya is there because "VFX Pipeline Integration" contains the alias pipeline integration. Backblaze qualifies by storing build artifacts, and Slack by sending notifications.
One member isn't a company at all. Evals, with 20 APIs, describes itself as a landscape catalog, and it's on the page because of an entry about wiring evals into CI. The post says it should not be counted as a market participant on any page.
Now the turn, and it's about the catalog's own record. The page's strong numbers, a median of 58.5 with 17 exemplars, come from what's on it. Of the 64, 21 are Amazon or AWS records. The purpose-built tools that did make it are spread out: GitHub Actions 82.8, GitLab 82.6, Bitbucket 64.2, TeamCity 52.1. The post's diagnosis is that a free-text field rewards whoever wrote the most sentences. A provider that describes itself as a CI server may not bother to write "CI" in a use-case list, because it's the whole product.
You might have copied the median into your slide as a sign of a healthy market. It's a sign of who filled in a field.
The post names the fix as its own: admit providers by tag as well as by use case, exclude catalog records, and tighten the pipeline integration alias. It doesn't say when that ships.
So you adjust your plan. You keep the page for discovery, since it did surface GitHub Actions, GitLab and Harness quickly. But you stop treating its order as a ranking of CI/CD vendors, at least until it's fixed. You open the records for CircleCI, Buildkite and the other left-out providers directly and read them yourself, which is what the post tells anyone choosing a pipeline tool to do. And you add one line to your spreadsheet: this page's membership depends on a free-text field, so absence from the list means nothing.
Read the original: https://apis.io/2026/11/03/ci-cd-page-ranks-salesforce-third-and-leaves-out-circleci-jenkins-and-buildkite/