WP_Discovery

An open agent-readiness and AI-discovery layer for WordPress.

WP_Discovery is an open convention for publishing a WordPress site's identity, capabilities, and APIs as a single machine-readable document at /.well-known/discovery.json — so AI agents, integrations, and crawlers can learn what a site is, what it exposes, and how to interact with it, without prior knowledge of which plugins are installed.

Plugins declare their capabilities (intent, in dot-notation such as commerce.products.read) through a registration hook. WP_Discovery collects, validates, normalizes, and de-duplicates those declarations into one document that consumers read at a predictable URL — they never need to probe individual plugins.

Plugin → Capability Declaration → WP_Discovery Registry → Normalized Discovery Model → /.well-known/discovery.json → AI / Agents / Systems

How it relates to llms.txt, Schema.org, MCP and A2A

Most AI-readiness signals describe the descriptive layer — who a site is and how its content may be used: an llms.txt summary, Schema.org, robots and content-usage directives. WP_Discovery covers that same ground and extends one layer further, into the operational: the same document also declares what a site can do — capabilities, APIs, A2A agent cards and MCP-shaped tools — so an agent can move from "I understand this site" to "I know which actions it exposes and how to call them." Transport protocols like MCP and A2A define how an agent calls a tool; WP_Discovery is the layer that declares which tools exist and points to them, built on the conventions the agent ecosystem is already converging on rather than a format of its own.

SignalDescribes the siteLets an agent act
llms.txtyes
Schema.orgyes
MCP / A2Ayes (transport)
WP_Discoveryyesyes — declares capabilities & points to MCP / A2A

Read the protocol

Machine artifacts

Project

FAQ

What is WP_Discovery?

An open convention for publishing a WordPress site's identity, capabilities, and APIs as a single machine-readable document at /.well-known/discovery.json, so AI agents and crawlers can understand and interact with the site without probing each plugin.

How is it different from llms.txt?

llms.txt is one descriptive signal — a text summary for language models. WP_Discovery includes that kind of descriptive signal but goes further: it also declares a site's capabilities and APIs, so an agent can act, not just read.

How does it relate to other AI-readiness efforts?

Most focus on the descriptive layer — signals that help AI understand who a site is, like an llms.txt summary, Schema.org, and content-usage directives. WP_Discovery covers that and adds the operational layer: declaring what a site can do. The approaches are complementary and can coexist.

Do I need a specific plugin?

No. WP_Discovery is an open spec any site or plugin can implement. Agentimus is the reference implementation for WordPress.

Is it an official standard?

It's an openly licensed, RFC-style specification with a public JSON Schema and a reference implementation — not (yet) an IETF, W3C, or WordPress standard. It's built on the conventions the agent ecosystem is converging on: the .well-known location, A2A agent cards, and MCP-shaped tools.