Agent-First CMS
Your agent
is your CMS.
One HTML attribute. Full content round-trip. Zero everything else.
<h1 data-adlib-cms="hero.headline">
Your Text Here
</h1>
Add it to your HTML. Run one command. Your agent handles the rest.
Four steps.
Three are automatic.
Annotate
Add data-adlib-cms attributes to your HTML. One attribute per content field. That's your whole schema.
Extract
Your agent reads the HTML, finds every annotation, produces structured content.json. Automatically.
Edit
Update the JSON. Change a headline, swap an image, add a testimonial. Plain data, no API calls.
Build
JSON → HTML. Push. Ship. Deterministic — same content in, same page out.
Zero everything.
No frameworks. No npm install. Just Node and your HTML.
No YAML. No env files. The HTML is the config.
Content lives in a JSON file. Version it with git.
No CMS UI to build. Your agent is the interface.
Remove the attributes. You still have a working site.
If you can write an HTML attribute, you already know adlib.
Built for agents.
Not dashboards.
Traditional CMSs were built for humans clicking buttons in browsers. Adlib was built for agents reading and writing files. Your client's content already lives in the HTML they shipped. Annotate it once and let your agent handle the rest — extraction, editing, rebuilding, deploying. No middleware. No APIs. Just files.
Ship it tonight.
Clone. Annotate. Extract. Push. That's the whole thing.
$ node extract.js index.html > content.json
→ 23 fields extracted. Done.