Quick example of how that works. Instead of the LLM rediscovering things every session, collab on creating a skill for specific task: get stories on current page, get meta, get comments.
Next time, the LLM gets told about skill on domain change, reads, executes efficientl.y
@zeeg I have multiple apps running, I was running a spotlight instance for each of them and used vite plugins to route errors to correct ports, how do I do that now? is there a `SENTRY_SPOTLIGHT_PORT` env var?