Unzip it into a folder you will keep — for example Documents/Ompera. Chrome runs the extension from this folder, so don't delete it afterwards.
You have a folder that contains manifest.json.
Load it in Chrome
Paste this into the address bar:
chrome://extensions
Turn on Developer mode (top right), click Load unpacked and choose the unzipped folder.
“Ompera Browser Connector” appears in the list, switched on.
Pin it (optional)
Click the puzzle icon in Chrome's toolbar and pin Ompera Browser Connector. Its icon shows whether your model is being shared and has a Stop button for all connections.
Connect your model
Open the connector page in the same Chrome profile. It detects the extension and looks for your model server.
Until the extension is published in the Chrome Web Store, Chrome installs it this way. The download is the same code we publish with a checksum. Developer mode only allows extensions you load yourself.
Update to a new version
Manually installed extensions don't update themselves. The connector page tells you when a new version is available or required.
Download the new version and unzip it over the same folder (replace the files).
In chrome://extensions, click the reload arrow on the Ompera card.
Reload the connector tab.
If you load a different folder instead, remove the old copy so only one Ompera extension is enabled.
What the extension can and cannot do
It can reach http://127.0.0.1 and http://localhost — your own computer — and only when the Ompera connector page asks.
It runs only on the Ompera connector page and cannot read other websites.
It remembers one setting: whether you pressed Stop. It never stores keys, tokens or request content.
To find your model server it checks a few well-known local ports (LM Studio, Ollama, llama.cpp and similar) with a read-only request.
While the connector tab shares a model, it asks Chrome not to unload that tab to save memory (from version 0.4.3).
