Back to directory

Browserbase Skills

Browserbase's official collection of Agent Skills that let Claude Code drive real or cloud browser sessions, capture traces, generate WebMCP tools and test web applications.

Browserbase Skills

Browserbase Skills is a collection of Agent Skills that let Claude Code work with Browserbase through browser automation and its official browse command-line tool. Individual skills cover automating a remote Browserbase browser session, including identity, verified browsers, CAPTCHA solving and residential proxies; deploying serverless browser automation to the Browserbase cloud; capturing a full DevTools-protocol trace alongside any automation run and splitting it into searchable per-page buckets; and turning a site's observed HTTP traffic into a best-effort OpenAPI specification.

Other skills include iteratively improving a browsing task until it reliably passes, optimizing an Agent API system prompt through repeated scored runs, building a local browser agent whose only capability is a domain-allowlisted CDP-gated tool, authoring and validating WebMCP tools for a specific site, syncing local Chrome cookies into a persistent Browserbase context, fetching static pages or web search results without a browser session, adversarial UI testing from a git diff, migrating browser-use automation to Stagehand, and auditing how agent-friendly a product's docs or SDK are.

It is written in JavaScript.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.