Browser
The page you are readingbecomes context.
Read the source you are citing without leaving the document, and attach it to your question when you want the agent to use it.
Free and open source. The agent sees a page only when you say so.
The page rides along with your question.
What you are reading is attached as one context item, so the answer is about that page and not the internet in general.
It asks before it looks.
Sharing a page with the agent is a decision you make each time, and you can say no.
Good to know
In the same window
A browser pane beside the document, not another app to switch to.
You allow each page
The agent reads browser content only after you allow it.
Your own tabs
Recent pages stay in the pane so you can come back to them.
Questions
- Is this a full browser?
- It is a pane for reading a page beside your writing, not a replacement for Chrome or Safari.
- Can the agent browse on its own?
- It can work with a page you have open once you allow it, and it asks each time.
- What does the agent see?
- The page content, and saved data such as cookies for that page, which is what the dialog warns about.
Keep the source next to the draft.
Free and open source.