WebMCP support now available in mcp-handler
mcp-handler 2.2.0 can expose selected MCP tools to in-browser agents through one script, proxying calls as the signed-in user without separate browser OAuth.
**mcp-handler 2.2.0** adds experimental **WebMCP** support. Developers opt tools into experimental_webMcp, then load a script from the MCP endpoint with the webmcp-script query parameter.
If you already maintain MCP tools, this offers a short path to reuse selected ones inside browser agents. The generated script registers them on the page and proxies calls to the existing server as the signed-in user.
**mcp-handler 2.2.0** adds experimental **WebMCP** support. Developers opt tools into experimental_webMcp, then load a script from the MCP endpoint with the webmcp-script query parameter. If you already maintain MCP tools, this offers a short path to reuse selected ones inside browser agents. The generated script registers them on the page and proxies calls to the existing server as the signed-in user. WebMCP remains a proposed standard and this integration is explicitly experimental. Tool exposure should therefore be narrow, with authentication and authorization checked at the server boundary.
This creates a direct reuse path from existing MCP servers into browser agents, reducing the need to rebuild selected tools for page contexts. Against the candidates, it advances discoverability beyond passive agent instructions but introduces a sharper authority boundary: calls execute through the existing server as the signed-in user. Because both WebMCP and the integration are experimental, narrow exposure and server-side authorization remain implementation requirements.