API Evangelist

Here is the strongest opinion I hold about agents and APIs right now, and I know it cuts against the mood of the moment: for most real integrations, you do not want the agent inferring what an API said at runtime through MCP. You want the agent to write deterministic code that integrates with the API, once, that you can read, test, and run a thousand times with identical behavior. The agent as a runtime interpreter of your APIs is a seductive idea and, for a lot of production work, a bad one. The agent as a very good developer who writes you a clean integration and then gets out of the way is the idea that actually holds up. https://lnkd.in/gPR-c-4v

In preparation for this week’s office hours I needed to summarize what matters most when it comes to forking Spectral. It helps me to go back through all of the conversations, issues, and roadmap items I have accumulated since announcing I was going to fork it, so that I can focus in on what actually matters. There is a lot more detail on the roadmap, but I wanted to go through it all and cherry-pick the things I want on the tip of my tongue. https://lnkd.in/gt4C7Jj6

Danny Preussler is an engineer on the API team at SoundCloud in Berlin, and he came up entirely on the client side — BlackBerry apps, then Android, then eBay, Groupon, and Viacom before landing at SoundCloud seven years ago. This is a conversation about one of the great API origin stories and one of the great API closures. SoundCloud shipped a public API in 2008, one year after the company started, and literally built its own product on the same API everyone else could use — the Android lead got hired because he wrote a better SoundCloud app on the public API than SoundCloud had. https://lnkd.in/g_57EkJ3

What is your vote to name the fork of Spectral? https://lnkd.in/gTWymb2m

You might have noticed the shift in my focus over the summer, but I want to be explicit about my departure from Naftiko for anyone who was away doing summer things, or who missed the signal in all the noise. After a year at Naftiko, I am leaving behind the startup I helped found to focus exclusively on API discovery with APIs IO and API governance with API Evangelist. The company and the product were not aligned with my vision for the API space, and as with all of my recent work, my vision is what propels me forward, not the paycheck or the venture capital investment. https://lnkd.in/gQQGtmUq

Mews tops the Food and Beverage listing on apis.io. 176 providers carry the tag; Mews leads with a composite Kin Score of 60.4/100 — strong — and 36.7/100 on agent readiness. The full listing, scored: https://lnkd.in/g4XkPk4T

Mintlify shipped mint signup, a CLI command that creates a Mintlify account from the terminal so a coding agent never has to open a browser to get started. Their reasoning is hard to argue with: they measured 250+ million agent queries across Mintlify-powered docs in July 2026, roughly 66% of their total measured traffic, and they name Claude Code, Cursor, and Devin as the clients they are building for. If two out of every three readers of your documentation is a machine, the front door probably should not be a signup form. https://lnkd.in/gMPc9BKd

The API Evangelist Rating System gives every provider two scores, and the second one is the one I care about most right now: agent readiness. It is a standalone number from zero to one hundred, and the most important design decision is that word–standalone. It is not blended into the composite quality score. A provider can be thin for humans and ready for agents, or strong for humans and useless to their agents, and collapsing those two situations into one number hides the exact gap the score exists to expose. So agent readiness rides on its own axis, and it answers a narrower question than composite quality: not “is this a good API,” but “can an autonomous agent drive this API without a human papering over the gaps?” https://lnkd.in/gRHy-jDT

Bentley Systems tops the Construction listing on apis.io. 283 providers carry the tag; Bentley Systems leads with a composite Kin Score of 64.1/100 — strong — and 24.8/100 on agent readiness. The full listing, scored: https://lnkd.in/gjtxTU25