The kit from
the room.

Use it when computer use / agent mode is on. It will download the ZIP, load the unpacked extension in Chrome, and pin it for you.

Install the ChatMultiAI Chrome extension for me using computer use. It is not in the Chrome Web Store, so you have to sideload it.

Do this in order, and stop if you get stuck:

1. Open this page: https://github.com/inspiretelapps/ChatMultiAI/releases/latest
2. Download the asset named ChatMultiAI-extension.zip from the latest release.
3. Unzip it into an easy folder, for example Downloads/ChatMultiAI-extension.
4. Open Google Chrome.
5. Go to chrome://extensions
6. Turn on Developer mode (top right).
7. Click Load unpacked.
8. Select the unzipped folder that contains manifest.json — not the ZIP file itself.
9. Confirm ChatMultiAI appears in the extensions list and is enabled.
10. Pin it to the Chrome toolbar: click the puzzle-piece Extensions icon, then pin ChatMultiAI.

If Chrome asks for confirmation, approve it. When you are done, tell me where the unzipped folder is and that the extension is pinned.

Paste this into ChatGPT or Claude. If it can use GitHub or a browser, it can do the setup. Otherwise it will walk you through the clicks.

Help me set up my own YouSummary website from this template:
https://github.com/inspiretelapps/vidbot-template

I'm not technical. Read the current README and setup files first, then help me get a working website. Use plain language and keep explanations short.

First, ask whether I already have a GitHub account. If not, explain briefly what GitHub is and guide me through creating a free personal account at https://github.com/signup, including email verification. I'll enter passwords and verification codes myself; don't ask me to share them in chat. Don't assume I know terms such as repository, Actions, Pages, or secret — explain each simply when needed.

Once my account is ready, ask me for my GitHub username, my YouTube playlist link, and what I'd like to name my library. Check whether you have tools that can access my GitHub account. If you do, carry out the setup steps your access allows. If you don't, guide me through one small step at a time with clear links and button names. Never claim you changed a setting or deployed something unless you verified it.

Create a new public repository in my account from the template, with an empty library. Work only in my new copy. I understand that the website, playlist links, and saved summaries will be public. If the name is already taken, ask before using an existing repository.

Help me obtain an OpenCode Zen API key and save it as the GitHub repository secret OPENCODE_API_KEY. Don't ask me to paste keys into this chat or into a file; direct me to the secure GitHub settings page. I'll handle sign-in, account creation, and any payment details myself.

Use DeepSeek V4.1 Flash unless I choose a Google Gemini option. Read the exact supported model IDs from the template's current update workflow. Use GitHub Pages for the website, with its publishing source set to GitHub Actions. Keep updates manual; no scheduled jobs or WhatsApp delivery.

Test one video from my playlist first. If YouTube captions fail, explain the issue simply and help me choose between adding a Supadata key as SUPADATA_API_KEY or uploading a transcript. Explain any extra service costs before I enable it. Don't invent a summary when captions are unavailable.

Run the first update when the required settings are ready. Check the run result, open the published website, and verify that a real summary and its chapter links are present. An empty website alone is not a completed test. If your tools cannot perform a step, tell me exactly what to click and continue once I've done it.

Finish by giving me my website link, my GitHub repository link, and a short explanation of how to update my library next time. Clearly state anything that still needs my attention.