Demonstrates the use of a LLM as a computer-use-agent (CUA) to control an embedded mobile web browser.
See the root readme.md for additional context. br
The default 'plan' is find 'neuro-symbolic ai' related content in twitter.
Alternate plans exist for Reddit and Amazon. They can be set to run via a minor code change.
Click the triangle icon to start the plan.
When the plan is started, the first 'task' waits for the user to login to the site. Its a manual step. Then click the continue icon >>| to let CUA drive the next steps. You may click >>| without logging in, if desired.
Depending on the plan selected, the plans collect some data from the various sites and call a tool to store that data.
Drop the burger menu and click the report (or people) icon to view the collected data.