Adapters

An adapter tells agentprose where posts come from. Without one, agentprose uses the zero-config source (auto-discovery) and detects WordPress on its own. With one, search, lists and full posts come straight from your CMS’s API.

CMS	Script tag	Safe in the page
WordPress	automatic, or data-adapter="wordpress"	nothing needed
Ghost	data-adapter="ghost" data-key="<content key>"	Content API key
Sanity	data-adapter="sanity" data-project="<id>"	no token (public dataset)
Contentful	data-adapter="contentful" data-space="<id>" data-token="<CDA token>"	Delivery API token
Enonic XP	data-adapter="enonic" data-endpoint="<guillotine url>"	public endpoint
Strapi 5/4	data-adapter="strapi" data-api="https://cms"	Public role, no token
Storyblok	data-adapter="storyblok" data-token="<public token>"	public access token
Any JSON	data-adapter="generic" data-src="/posts.json"	n/a

Each adapter is a separate 2–3 KB file that the script tag loads only when data-adapter asks for it. With npm, import it from agentprose/adapters/<name>. Every attribute per adapter is in the adapter reference.

The one rule: the page is public

WebMCP tools run in the visitor’s browser, so any credential in the page is public. Use only credentials the vendor designed to be public (the ones in the table above), or use proxy mode.

agentprose enforces this: credentials shaped like management or admin tokens (Contentful CFPAT-…, Sanity sk…, Ghost Admin id:secret) make the adapter throw instead of running.

Proxy mode

When content needs a secret (a private Sanity dataset, a Strapi API token, a preview token on staging), put a small proxy between the page and the CMS:

<script src="https://cdn.jsdelivr.net/npm/agentprose@0/dist/agentprose.min.js" defer
        data-adapter="contentful" data-space="xxxx" data-proxy="/api/cms"></script>

Each request keeps the vendor’s path and query but goes to your proxy, which allow-lists read-only paths, adds the credential and forwards the request. See Private content: proxy mode.

Your own adapter

Only list and get are required:

import { defineAdapter } from 'agentprose/adapters/generic';

export const myCms = defineAdapter({
  name: 'my-cms',
  async list({ limit = 20, offset = 0 } = {}) {
    const r = await fetch(`/api/posts?limit=${limit}&offset=${offset}`).then((r) => r.json());
    return { items: r.posts.map((p) => ({ url: p.url, title: p.title, summary: p.excerpt, published: p.date })), total: r.total };
  },
  async get({ slug, url }) {
    const p = await fetch(`/api/posts/${slug ?? new URL(url).pathname.split('/').pop()}`).then((r) => (r.ok ? r.json() : null));
    return p && { url: p.url, title: p.title, markdown: p.markdown };
  },
});

When search is missing, agentprose ranks list() results itself; when taxonomies is missing, it counts them from posts. Return markdown or html from get (HTML is converted). More in Writing an adapter.