About agent actions: browse, scraper, and search

[Pinucci0]

In LocalAI’s agent configuration, there is the option to select the following three “actions”:

• browse
• scraper
• search

The questions are:

1. What is the actual functional difference between "browse" and "scraper"?
2. With which HTTP browser headers are the "browse", "scraper", and "search" actions performed?
3. Which search engines does "search" query? Can it be configured to use other search engines?

I’m asking question n.2 because LocalAI was also designed to ensure privacy by running AI models locally. However, a potential privacy vulnerability with local models arises from internet-based searches. Therefore, it would be important—not only to allow users to choose which search engines to use—but also to enable customization of browser headers so that LocalAI’s web requests can be “disguised” as typical browser traffic, blending in among legitimate user requests.