Opera has introduced MCP Connector for Opera Neon, allowing external AI clients to connect directly to the browser.
The tool lets third-party AI agents work inside a user's active browser session rather than in a separate environment.
Opera Neon, the company's browser focused on AI-driven workflows, now supports clients including Claude, ChatGPT, Lovable, n8n and OpenClaw, along with other systems that support the Model Context Protocol, or MCP. Through that connection, AI tools can access live browser context, including open tabs, page content and authenticated sessions.
Connected AI clients can then carry out tasks inside the browser, such as navigating pages, extracting information, taking screenshots, filling in forms, opening new tabs and running searches.
The move addresses a common limitation of current AI assistants, which often require users to copy information from browser tabs into separate tools and restate what they want to do each time they switch applications. Opera is positioning the browser as the place where that context already exists.
"Last year, we launched Browser Operator as a first step toward an agentic browser. Now we are opening those capabilities to external AI clients through MCP, so they can act directly inside the browser, not outside it," said Monika Kurczyńska, Director of R&D for Browser AI at Opera.
Open access
The announcement marks a broader effort to build an open connection between browsers and external AI systems. Rather than limiting these functions to its own built-in tools, Opera is exposing an MCP endpoint that outside clients can use to interact with the browser while the user is working.
This differs from AI systems that rely on isolated or simulated browser environments. In Opera Neon, the AI client works within the user's actual browser session, where tabs, log-ins and live page states are already present.
The approach could appeal to developers and automation users who want AI tools to interact with websites and web applications in real time. Developers are already using tools such as Claude Code to test applications in a live browser environment, while design and prototyping software such as Lovable can work with live interfaces. Automation platforms including n8n and general AI assistants such as ChatGPT can also add browser actions to their workflows.
Technical layer
To support the connector, Opera has built two main technical components. Authentication runs through a secure MCP server URL designed to ensure that only authorised AI clients can access the browser session.
A separate proxy layer is intended to keep the connection stable between the browser and the AI tool. It also returns a clear "browser not available" state when the browser cannot be reached.
Opera has been developing several browser products for different user groups, including Opera One as its flagship browser, Opera GX for gamers and Opera Neon for AI-led workflows. MCP Connector extends Opera Neon's existing task-execution functions by making them available to external AI systems, rather than only internal browser features.
That reflects a wider shift in the browser market as software companies test how AI agents can complete actions on behalf of users, not just respond to text prompts. Browser-based AI has become an area of competition because web browsers remain the main interface for work, shopping, communication and software access.
MCP Connector is available to all Opera Neon subscribers. Opera also plans to bring a simpler version of the browser connector to Opera One and Opera GX, extending the same type of access across more of its browser portfolio.
Kurczyńska said the aim is to connect AI tools to where users already do their digital work. "The browser is where workflows live, but AI has been disconnected from it," she said. "With Opera Neon, we connect popular AI clients directly to an agentic browser, so they can operate where users already work, without needing to recreate context."