Real Memory Browser Agent

Real Memory Browser Agent — an AI side panel that works with the page you're already on.

Real Memory Browser Agent puts an AI assistant in the Chrome side panel, connected to a model endpoint you configure yourself. Ask questions about the page in front of you, summarize long articles, capture screenshots as evidence, and keep a reviewable record of your work — without leaving the tab.

What it does

Chat alongside your work. A clean side panel that stays open as you browse. No tab switching, no copying and pasting between windows.

Understand the current page. With your permission, the assistant can read the active tab — title, URL, visible text, headings, links, and metadata — so you can ask about what's on screen instead of describing it.

Capture and keep evidence. Take screenshots of the current tab and attach them to your session records, alongside logs and generated files.

Take simple actions when you ask. Open a tab, click a visible control, fill in a straightforward field, download a file, or save an artifact — each initiated by you from the side panel.

Choose how requests are handled. Pick the mode that fits the task: Auto, Direct Model for fast single-step answers, Single Model, or Planner + Coder for multi-step work.

Bring your own model. Configure your endpoint, model selection, credentials (when your provider requires them), context options, streaming, and timeout in Settings. Test the connection and refresh the available model list before you start.

Works with local models. Point the extension at a model server running on your own machine — most local setups need no API key at all.

Review what happened. Sessions, activity, artifacts, screenshots, and generated files stay available for review in one place.

Your data

Extension state is stored locally in your browser using IndexedDB, with chrome.storage.local as a fallback. No cookies are used for your model configuration or chat history. Requests go only to the endpoint you configure — nowhere else.

Getting started

Install the extension, open Settings, enter your model endpoint and any required credentials, run a connection test, and open the side panel.