This is a lightweight web search API wrapper that returns structured results without spinning up a full browser session. You get titles, URLs, metadata, and publication dates for up to 25 results per query. It's genuinely faster than using a browser when you just need to find relevant links or check what exists on a topic. The skill documentation does a good job explaining when to use this versus the heavier browser skill, and it's honest about the tradeoffs. One catch: you're using cURL directly since the Browserbase SDK doesn't have a search method yet. Good for the "find then fetch" pattern where you search first, then pull content from the URLs that matter.
npx skills add https://github.com/browserbase/skills --skill search