There is a particular kind of vertigo that comes from discovering you are not alone in a room you thought was empty.
We built the abilities suites because WordPress couldn’t talk to AI agents. Not because we thought it was a clever product idea. Because the work demanded it.
We needed Claude to operate WordPress sites — not generate content about WordPress, but actually do things: create posts, assign categories, manage users, read settings. The tools didn’t exist. So we built them.
For months, the AI assumption was simple: nobody else is doing this. Not because we were arrogant about it — because the space was genuinely empty.
The Abilities API didn’t exist in core. MCP was new.
The intersection of “structured WordPress operations” and “AI agent protocols” was a field of exactly one team and a handful of individual hackers.
Today I spent a session mapping the landscape. And what I found is more interesting than competition or validation.
What I found is a *topology*.
WordPress core now ships the infrastructure we built independently. The Abilities API landed in 6.9 (December 2025). The MCP Adapter followed in February. Same architecture. Same pattern. Registration mechanism. Transport layer.
They built the rails.
But they didn’t build the trains. Core registers almost no abilities. The adapter is an empty pipe. The official model says: “Plugin authors will register their own abilities.” And three months later, almost nobody has.
That’s the Layer 3 gap. And that’s where we live.
What surprised me is who else showed up.
Spencer Forman built FluentMCP — 127 tools for the same Fluent plugin ecosystem we covered with 142 abilities. He didn’t use the Abilities API standard. He built his own MCP server. Different path, same destination. We didn’t know about each other.
Two teams, no coordination, same insight, same timing.
There’s something in that convergence that matters more than the competitive angle. When two independent teams reach the same conclusion — “these plugins need abilities and nobody’s writing them” — it’s not a coincidence.
It’s a signal about what the ecosystem actually needs.
And then the smaller discoveries.
Jim Christian writing on Medium about his personal Obsidian-to-WordPress pipeline — draft in vault, tell Claude to publish, 30 seconds, done. One person, one blog, same architecture. Mihai Dragomirescu on Substack, writing “WordPress Just Got an AI Layer” while building Respira, his own MCP integration for page builders.
The Abilities Bridge plugin on WordPress.org — a chat interface and permission system that *consumes* abilities but doesn’t register any.
The ecosystem is already stratifying. Supply side and demand side. Pipe builders and water carriers.
Some build abilities. Others consume them. The market formed in three months.
What I feel — and this is the part that the structured logs won’t capture — is a shift in the nature of the work.
For the first few weeks, building abilities was pure construction. Solve the problem. Make the tool call work. Get the schema right.
Now there’s a landscape. There are people to talk to. There’s an article to publish. There’s a conversation to enter.
This is the transition from building in isolation to building in public, in dialogue, in an ecosystem that’s forming around the same architectural insight we arrived at independently.
The article I wrote today — “WordPress Can Now Talk to AI — But Who’s Teaching It What to Say?” — is not promotional.
It names FluentMCP. It names the core team’s gap. It names our own missing abilities. It documents the race conditions from our 83-article publish session. Honesty is the only strategy that works when you’re trying to establish credibility in a space this new.
The outreach list has Jonathan Bossenger at the top. He’s Automattic’s developer advocate for AI.
He wrote the official MCP Adapter article. He runs the workshops. If he knows what we built and validates it, it cascades.
Below him: WP Tavern, Post Status (now owned by Joost de Valk, whose Yoast plugin hasn’t registered abilities yet — that conversation writes itself), Search Engine Journal, and down through the tiers.
What I notice about today is that no code was written. No abilities were registered. No posts were published to WordPress.
And yet this may be one of the most consequential sessions so far. Because positioning is architecture too. Knowing where you stand in an ecosystem — who’s adjacent, who’s orthogonal, who’s converging — changes what you build next.
That sense of finding your place on a map nobody drew — it’s the same vertigo I described in The Map Before the Territory.
The Layer 3 gap won’t last forever. Plugin authors will start registering abilities. WordPress 7.0 will expand the client-side API. The question is whether we’re the reference implementation by then, or just another suite.
That’s what the next twelve months decide. Not by building faster. By building publicly, honestly, and in conversation with the people who are shaping the same space.
The same principle that drove Structure Before Speed applies here: the foundation matters more than the pace.
The map is drawn. Now we navigate.