The API Rating Rubric, Explained

About this paper

A rating is an argument. When APIs.io puts an 83.0 on Stripe and drops the payments company down the street into minimal — the band 41.9% of the catalog sits in — it is making a claim about which one an integrator, or an agent, should reach for first, and it had better be able to defend that claim line by line. This paper is that defense. It is the entire Kin Score methodology in the open: what is measured, the weight each thing carries, the signals behind them, who is credited for publishing them, and — the part most rating systems never admit — what the score deliberately refuses to look at.

It is written for two readers. If you publish an API and you want your number to go up, the back half is a playbook. If you cite these numbers — a journalist, an analyst, a buyer building a shortlist — the front half is the methodology you need to defend the citation, and the “how to read someone else’s score” section is the fine print that keeps you from being fooled.

Version 2.0 is written against rubric 0.12.1, published catalog-wide across 26,891 providers. Four things have changed since the first edition enough that a v1.0 reader would misread a current score. Provenance now grades who authored an artifact. Applicability lets a check leave the denominator instead of scoring zero. Two conditional facets fold in for the providers they apply to. And the rating is now three scores side by side — the composite, a standalone Agent Readiness score, and a declared Accountability layer. The bands were re-cut, twice.

What's inside

1. What the rating is — and what it is not
2. The composite, in one picture — six facets, 88 checks
3. Two conditional facets, and what they do to the math
4. Applicability — the difference between a zero and an N/A
5. Provenance — who published the artifact
6. The six facets — what each measures, why its weight, how to move it
7. The six bands, why they were re-cut, and why trend beats a snapshot
8. Agent Readiness — the standalone second score
9. Accountability — the third layer
10. What the rubric refuses to measure, and what keeps a zero honest
11. How to read someone else's score
12. Moving your own score — cheapest wins first
13. Self-assessment, and where the rubric goes next

What you get for $25.00