Install guide

Install the latest build and send your first compare prompt.

This is the shortest path to a real product trial. You do not need to build the project locally to evaluate MultiAiPanel.

Before you start, keep the supported AI tabs you want to compare already open and signed in inside the same browser profile where you load the extension.

Once the side panel opens, use the built-in first-run checklist to open tabs, refresh readiness, and seed the first compare prompt without bouncing back into the docs.

If you want the shortest operator-style map from “extension installed” to “first compare run really succeeded,” use the first compare guide. It keeps the first-run checklist, repair ladder, and login-gated explanation in one place.

Keep the order honest: browser install first, one compare second, MCP or Docker builder setup only after the core compare-first path is already working.

The supported install path today is the packaged GitHub Release zip. Browser-store submission materials are being kept ready, but GitHub Releases remains the supported install surface today.

01

Load the extension

Use the packaged release zip and skip local source setup on the first pass.

02

Run one compare

Keep your trusted AI tabs signed in, open the side panel, and use the built-in checklist before sending.

03

Keep the turn alive

Retry only the failures, export a readable artifact, or move to prompt packs before touching builder docs.

What you need first

  • A Chromium-compatible browser with Developer Mode available
  • At least one supported AI chat tab already open and signed in
  • A local folder where you can unzip the packaged release

Quick install path

  1. Download the latest packaged release zip and unzip it locally.
  2. Open chrome://extensions, enable Developer Mode, and use Load unpacked on the extracted folder.
  3. If the MultiAiPanel icon is hidden, open the browser Extensions menu, pin it, then click the toolbar icon to open the side panel.
  4. Open your supported AI tabs, then ask once from the side panel.

Current distribution surface

Today the supported path is the packaged release zip from GitHub Releases. A lower-friction browser-store path is being prepared, but it is not live yet.

What you do next

  • Run one real compare turn from the side panel.
  • Retry only the failed models instead of replaying everything.
  • Export a readable compare summary or Markdown artifact.
  • Reuse prompt packs for the second run instead of starting blank.

Real Chrome proof note

For day-to-day automated runtime proof, prefer Chromium or Chrome for Testing. Official Google Chrome branded builds 137+ / 139+ no longer reliably auto-load unpacked extensions from command-line flags.

If you are validating MultiAiPanel in a real signed-in Chrome profile, keep that profile open, go to chrome://extensions, enable Developer Mode, and use Load unpacked on the extracted release folder. The real Chrome lane is a manual proof lane, not the default automated runtime path.

After the first compare works

Stay in the product lane first: copy a compare summary, export Markdown, retry only the failed cards, or open prompt packs for the next run.

Optional AI analysis is available too, but it still runs through a browser tab you already trust. The default compare workflow does not require any hosted relay or separate account layer.

If you are here for the coding-host route, open the MCP agents page and Docker integration page only after the product path is clear. The repo-owned MCP server is secondary, not required for the first compare.

Need the maintainer workflow instead?

Local builds, release bundles, front-door checks, asset refresh steps, and the repo-local mcp:operator helper live in CONTRIBUTING.md so the public front door can stay focused on first-run evaluation.

Need more public-facing context before you install? Read the supported sites, trust boundary, and FAQ pages.