what is livewire-webmcp fr?

livewire-webmcp is a small package for Laravel and Livewire applications that lets you expose specific actions in your app as tools an AI agent can call directly from the browser, using an emerging standard called WebMCP. In practice, this means you can mark a Livewire method with an attribute, and if the visitor's browser supports the underlying WebMCP feature, an AI assistant running in that browser can discover and call that method as a structured tool, alongside the page working normally for everyone else. It supports Laravel versions 10 through 13 and both Livewire 3 and Livewire 4. Because WebMCP is described as progressive enhancement, browsers that do not yet support it simply see the Livewire component behave as it always has, with no broken functionality for those visitors. Setting it up involves installing the package through Composer, publishing its front end assets, and adding one line to your Blade layout right after Livewire's own scripts are loaded. If your browser requires a Chrome origin trial token for the underlying WebMCP feature, there is a configuration setting available for that as well. To turn a Livewire action into a tool, you add a WebMcpTool attribute above the method, giving it a title, a description, and a JSON schema describing what input it expects, plus a list of which of those inputs map to the method's actual arguments in order. Each mounted instance of a component gets its own uniquely named tool in the browser, so multiple copies of a form on one page stay distinct. The README is direct that marking a method this way does not automatically make it safe. You are still expected to validate input, check permissions, and avoid returning sensitive data, exactly as you would for any other action a user could trigger, and any return value from the method becomes visible to whatever agent called it. Tools are treated as not read only by default, so you have to explicitly mark an action as read only if it truly has no side effects on the application.