ToolsAugust 4, 2026via Vercel Blog
Give your eve agent a browser
Why it matters
A major capability expansion for agentic AI frameworks: agents can now interact with web interfaces autonomously, expanding the scope of tasks they can handle without human intervention. This moves agent tooling closer to production-ready web automation.
Key signals
- Vercel releases .agent-browser extension for eve agents
- Agents gain full browser capabilities: navigate, read, click, fill forms, screenshot, inspect console/network
- Runs in agent sandbox, not app environment
- Namespaced tools include browser__navigate, browser__snapshot, browser__click, browser__fill, browser__screenshot
- Includes domain allowlist, credential protection, and sandbox isolation controls
- Cookie/storage/auth commands never exposed to model
- Documentation and Next.js example provided
- Published August 4, 2026
- Vercel Eve framework adds .agent-browser extension for agent web navigation
- Agent tools include: browser__navigate, browser__snapshot, browser__click, browser__fill, browser__screenshot
- Runs in sandboxed environment separate from application code
- Sandbox security includes domain allowlist, credential protection (cookies/auth never exposed), and tool override controls
- Snapshot refs enable agents to inspect and act on specific page elements
- Published August 4, 2026; part of broader agent framework maturation
The hook
Vercel's eve agents can now browse the web natively—navigate, click, fill forms, take screenshots, all sandboxed and credential-safe.
Your eve agent can now navigate the web like a human with .agent-browser
The extension gives any agent a full set of browser tools: navigate pages, read content, click, fill forms, take screenshots, and inspect console and network activity. Everything runs inside the agent's sandbox.@agent-browser…