lonkedIn

Do you sometimes wish there was less waffle and BS on LinkedIn?

Me too, so I made a Chrome browser extension that reads your feed and uses a small local AI model to give you a close friend's, mildly sarcastic, summary for each post instead. Then just click it to read the full post, if it's right for you.

lonkedIn: "We have LinkedIn at home."

Just a quick experiment, completely for fun, and provided as is.

How To Install And Activate

Option 1: Use the ZIP

1. Download the latest release: https://github.com/slicedPear/lonkedIn/releases/latest
2. Unzip it somewhere sensible.
3. Open Chrome, Edge, Brave, or another Chromium browser.
4. Go to chrome://extensions.
5. Turn on Developer mode.
6. Click Load unpacked.
7. Select the unzipped folder that contains manifest.json.
8. Find lonkedIn in your extensions list.
9. Click the puzzle-piece icon in the browser toolbar and pin lonkedIn so it is easy to reach.

The GitHub Code > Download ZIP button downloads the source code, not the ready-to-load extension. Testers who just want to install and go should use the release ZIP above.

Option 2: Build It From Git

npm install
npm run build

Then go to chrome://extensions, turn on Developer mode, click Load unpacked, and select the dist/ folder.

Do not select the repository root folder. Chrome needs manifest.json at the top level of the folder you load, and the build puts that file in dist/manifest.json.

How To Use

1. Open LinkedIn and go to your feed.
2. Click the pinned lonkedIn extension icon.
3. Switch it on.
4. Choose a local model.
5. Wait for the model to download the first time.
6. Refresh LinkedIn if needed.
7. Give it a few seconds after the feed loads. It has to notice the posts, queue them up, and let the tiny local brain do its thing.
8. Click any lonkedIn summary in the feed to toggle back to the full original post.

You can swap models at any time from the extension popup. The new model will download the first time you choose it, then should be reused from the browser cache after that.

Model Options

The models come from Hugging Face and run locally in your browser using WebLLM/WebGPU. The model data downloads on first use; the extension does not send your LinkedIn posts to a remote AI service.

Size	Model	Approx Download	Source
Small	gemma3-1b-it-q4f16_1-MLC	0.7 GB	mlc-ai/gemma3-1b-it-q4f16_1-MLC on Hugging Face
Medium	gemma-2-2b-it-q4f16_1-MLC	1.9 GB	mlc-ai/gemma-2-2b-it-q4f16_1-MLC on Hugging Face
Large	gemma-2-9b-it-q4f16_1-MLC	6.3 GB	mlc-ai/gemma-2-9b-it-q4f16_1-MLC on Hugging Face

Start with Small if you just want to see it work. Try Medium if your machine has a bit more headroom. Large is there for people whose laptops sound calm under pressure.

Notes

• Requires a Chromium browser with WebGPU support.
• First model download can take a while because the model files are chunky.
• After the model is cached, normal use should be faster.
• This plugin is a light-hearted look at LinkedIn culture and should not be taken seriously.
• This is source-available for noncommercial use under the PolyForm Noncommercial License 1.0.0.

Privacy

lonkedIn is local-first:

• no backend
• no telemetry
• no remote inference API
• no LinkedIn post text leaving your device for summarisation

The only expected network activity is downloading model data from Hugging Face.