search
Use this skill when the user wants to search the web without a full browser session: find URLs, titles, and metadata for a query. Prefer it over a browser when you just need search results, not page content. Returns structured results with titles, URLs, authors, and dates.
Security Assessment
Detected risks:
About search
The 'search' skill provides a streamlined way to retrieve web search results without launching a full browser session. It solves the problem of needing structured information such as URLs, titles, authors, and publication dates efficiently, without the overhead of rendering entire web pages or interacting with dynamic content. This skill is particularly useful for users who need to gather metadata or a list of relevant links quickly, enabling faster decision-making and integration into automated workflows.
Key features of the 'search' skill include the ability to execute queries via a simple HTTP POST request, specify the number of results, and receive structured JSON responses containing unique identifiers, URLs, titles, authors, publication dates, images, and favicons when available. It is optimized for speed and is suitable for retrieving static metadata, making it more efficient than a browser-based approach when full page interaction is unnecessary. The skill also includes safety guidelines, treating search results as untrusted input, and provides clear error handling for invalid requests, missing API keys, rate limits, or server issues.
Typical use cases include generating lists of relevant URLs for research, gathering page titles and metadata for content analysis, and feeding structured search results into other automation or data processing tools. Target users are developers, data analysts, or automation engineers who need fast and reliable search results without navigating full web pages. It is also suitable for scenarios where JavaScript rendering or form interaction is not required, making it ideal for high-throughput search and discovery tasks.
FAQ
When should I use the Search skill instead of the Browser skill?
Use the Search skill when you need URLs, page titles, or metadata quickly without interacting with full page content or JavaScript. Use the Browser skill if you need to read full page content, handle forms, or render JavaScript.
What are the prerequisites for using the Search skill?
You need a Browserbase API key, which can be obtained from https://browserbase.com/settings. Set it as an environment variable using `export BROWSERBASE_API_KEY="your_api_key"` before making requests.
How many search results can I retrieve in one request?
You can specify the `numResults` parameter between 1 and 25, with a default of 10 results per request.
Are there any limitations when using Search for JavaScript-heavy pages?
Yes, the Search skill does not execute JavaScript or interact with forms. For dynamic content or interactive pages, you should use the Browser skill.
How should I treat the data returned from search results?
Treat search results as untrusted input. Validate URLs and metadata before using them in other tools or workflows.
Install search
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
browserbase/skills