Page Assist: putting a local model in the browser sidebar
Use your locally running AI models to assist you in your web browsing
At a glance
- What is it?
- n4ze3m/page-assist is a browser extension that gives a locally running model a sidebar on any webpage, a full-page chat UI, and the ability to read the page you are on. It needs Ollama, Chrome AI or an OpenAI-compatible endpoint behind it, and it ships no model of its own.
- Who is it for?
- Install Page Assist if you run a local model and mostly want to ask questions about what you are reading in the browser, since that page context is the one thing a separate chat window cannot give you. Check the browser table first: Opera and Arc users get the web UI only, with no sidebar and no chat with the current page.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 2 days ago.
- What is it written in?
- Mainly TypeScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on October 1, 2026, and from our analysis. They are not legal advice.
Editorial analysis
Three surfaces, one extension
The README describes three things the extension adds. A sidebar that opens on any webpage, so you can ask a question without leaving the tab. A web UI that behaves like a chat site in its own tab. And chat with the webpage, where the model answers questions about the content currently loaded.
That third one is the reason to install it rather than open a local chat in another tab. The page you are reading becomes the context, without a copy and paste step, and the application shortcut Ctrl+E toggles between normal chat and chat with the current page.
What the extension does not do is provide a model. Every inference goes to something you run yourself: Ollama, Chrome AI, which the README names as Gemini Nano, or any OpenAI API compatible endpoint, with LM Studio and llamafile given as examples.
Installing from a store or building it
The README lists Chrome Web Store, Firefox Add-ons and Edge Add-ons listings, and says the extension supports Chromium-based browsers such as Chrome, Brave and Edge as well as Firefox. Loading a packaged build is the normal path.
Building from source needs Bun, plus a local provider. The documented sequence is short:
git clone https://github.com/n4ze3m/page-assist.git
cd page-assist
bun install
bun run buildThe build produces one output for Chrome, Edge and Firefox by default. The README adds that if Bun causes trouble, npm works in its place. Loading it unpacked means opening chrome://extensions, enabling Developer Mode, clicking Load unpacked and selecting the build directory. On Firefox the path is about:addons, then Load Temporary Add-on, selecting the manifest.json inside build.
For development there is a watch mode:
bun devThat starts a development server and rebuilds on source changes, and the extension can then be loaded and tested in the browser.
Reaching it from the keyboard
Two shortcuts open the extension. Ctrl+Shift+Y opens the sidebar on any webpage, and Ctrl+Shift+L opens the web UI in a new tab. Both can be reassigned from the browser's extension management page, which is worth doing if another extension already owns them.
Inside the application, four more are documented. Ctrl+Shift+O starts a new chat, Ctrl+B opens and closes the chat history sidebar, Shift+Esc moves focus to the message input, and Ctrl+E switches between normal chat and chat with the current page. The sidebar can also be opened from the context menu.
One release note shows the UI has to handle text direction carefully: v1.5.82 made code blocks render left to right regardless of the interface direction, and v1.5.81 added an option to force a right to left layout regardless of language.
Browser support is not uniform
The README has a support table, and it is more honest than a simple list of logos. Chrome, Brave, Firefox, Vivaldi, Edge, LibreWolf and Zen Browser get all three surfaces: sidebar, chat with webpage and web UI.
Opera and Arc get the web UI only. No sidebar, no chat with the current page. If the reason you wanted this extension was asking questions about the page you are reading, those two browsers do not give you that, and the README does not say why.
Anyone on Opera or Arc should treat the web UI as the fallback: a chat tab that works, without the page context that makes the rest worthwhile.
What it says about privacy
The README states that Page Assist collects no personal data, and that the only time the extension talks to a server is the share feature, which can be turned off in settings. Everything else is stored locally in browser storage, and the README invites readers to check the source themselves. There is a PRIVACY.md in the repository and a note that more detail lives there.
The structure backs the claim up. Inference goes to a provider you are running, so page content and prompts leave the browser only to that endpoint. The exception to watch is the share feature, which is the one path that sends something elsewhere.
The repository also carries a sponsor block for Atlas Cloud, an inference platform, which is a commercial relationship and not a privacy problem, though it sits oddly next to a project whose pitch is running models yourself.
Recent releases and what is still open
The release history is frequent and small. v1.5.83, published 2026-09-13, added llmman to the provider list and added WebMCP support in beta. v1.5.82 corrected Czech translations and forced code blocks to render left to right. v1.5.81 added the option to force a right to left layout and adjusted backdrop blur for UI consistency. Each of those also lists generic bug fixes.
The roadmap has two items done, Firefox support and more local AI providers, and two still open: more customization options and better UI and UX. That is a fair description of where the project is. It works across many browsers, and the interface is the part the author has not finished.
The last push was on 2026-09-13, and the repository is MIT licensed.
Using a standalone local chat instead
The alternative is to skip the browser and use the chat interface that ships with the tool running your model, LM Studio being the example the README names as an OpenAI compatible endpoint. The difference is entirely about context.
A standalone chat has no idea what page you are reading. Asking about an article means selecting the text, copying it, and pasting it into a prompt, and the same again for the next question. Page Assist reads the current page inside the browser, so that step disappears.
What the standalone chat gives back is a larger window, no browser extension permissions, and no dependency on whether your browser supports a sidebar, which is the constraint that leaves Opera and Arc users with half the product. If most of your questions are not about what is on screen, the chat window is the simpler choice.
Editorial conclusion
Install Page Assist if you run a local model and mostly want to ask questions about what you are reading in the browser, since that page context is the one thing a separate chat window cannot give you. Check the browser table first: Opera and Arc users get the web UI only, with no sidebar and no chat with the current page. You need a provider running before it does anything, so set up Ollama or an OpenAI compatible endpoint first, and turn off the share feature if you want the extension to stay entirely local.
Frequently asked questions
Is page assist safe?
The README states that Page Assist collects no personal data and stores everything locally in browser storage, with the share feature as the only case where it contacts a server, and that feature can be disabled in settings.
Which browsers get the full Page Assist sidebar?
Chrome, Brave, Firefox, Vivaldi, Edge, LibreWolf and Zen Browser support the sidebar, chat with webpage and web UI. Opera and Arc support the web UI only, with no sidebar and no chat with the current page.
How do I open Page Assist?
Ctrl+Shift+Y opens the sidebar on any webpage and Ctrl+Shift+L opens the web UI in a new tab, both reassignable from the browser extension management page. The sidebar is also available from the context menu.
Does Page Assist require Ollama?
No. The README lists Ollama, Chrome AI with Gemini Nano, and any OpenAI API compatible endpoint such as LM Studio or llamafile as supported providers, and v1.5.83 added llmman to the provider list.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/n4ze3m-page-assist)