How does the APIs.io rating system work?

[kinlane]

This is a thread to discuss the ratings engine behind APIs.io, providing a forum for defining, iterating, and improving upon what is currently used as the ratings applied to APIs, which are then sorted by as part of the search.

Here is a dump of the concepts at play:

• Uses Spectral - The rating system is based upon Spectral rules that have a score applied to them, which then get applied to any OpenAPI the rules are run against.
• Rules - These are the rules we currently have defined with the points applied to each--the goal is to keep adding and refining the rules until we end up with the right algorithm collectively.
• APIs.json - All rules are developed for and applied to APIs.json evaluating the metadata, but also the APIs and their properties, as well as the common properties across an index.
• Properties - All of the properties present for an API, indexed as part of it's APIs.json are pulled, and evaluated based upon whether a 200 or 404 is returned, and further processing of other formats like OpenAPI.
• OpenAPI - When an API has an OpenAPI indexed, more rules can be applied to the OpenAPI, adding even more points to an API based upon the completeness of the OpenAPI.
• APIs.io Detail - Every API published to APIs.io has a list of rules it has complied with, which are added up.

I will write up a more refined blog post about the rating system, as well as publish a page to APIs.io to support the rules page. However, i wanted to start a general discussion about how the rating system is being applied behind APIs.io, so we can ultimately move forward as a community.

[philsturgeon]

Very cool use of Spectral. One cool roadmap item could be dinging the score based on security issues found by the spectral owasp ruleset.

https://github.com/stoplightio/spectral-owasp-ruleset

[kinlane]

Thanks! I will add in and group so I can report upon.