Skill does not use agent-browser session by default, causing issues when parallel agents are testing #214
Description
If you install the claude skill of agent-browser and attempts to do many tests at the same time in parallel on your machine, they overlap with 1 browser. Shouldn't we update the skill to use agent-browser session each time it starts a new test or task that involves the browser to mitigate that?