feat: 浏览器一期：坐标点击 + live 截图

- browser_click_at(x, y)：视口坐标点击（1280x800、左上角
  原点），越界与非数字拒绝；live 视图里用户在截图上的
  点击落点就走这个工具。
- browser_live_shot：固定文件 browser-live.png + 状态
  sidecar browser-live.json（URL/标题/时间戳），watcher
  轮询一个稳定路径即可；技能要求每次动作后都调一次，
  保持画面最新。
- MCP 新增 browser_click_at / browser_live_shot 两个工具；
  技能补 live 视图一节；测试 9 项（含 fake mouse、越界、
  live 文件与状态读回）。

cloud/playwright_mcp.py

@@ -109,6 +109,44 @@ _TOOLS = [
             "required": ["expression"],
         },
     },
+    {
+        "name": "browser_click_at",
+        "description": (
+            "Click at viewport coordinates (CSS pixels, origin "
+            "top-left of the 1280x800 viewport). This is the tool "
+            "a live viewer drives: the user clicks on a screenshot "
+            "and the coordinates arrive here. Then snapshot to see "
+            "the result."
+        ),
+        "inputSchema": {
+            "type": "object",
+            "properties": {
+                "x": {"type": "number"},
+                "y": {"type": "number"},
+            },
+            "required": ["x", "y"],
+        },
+    },
+    {
+        "name": "browser_live_shot",
+        "description": (
+            "Screenshot to the FIXED live-view file "
+            "browser-live.png plus a browser-live.json status "
+            "sidecar (url, title, timestamp), so a watcher can "
+            "poll one stable path. Call after every navigate / "
+            "click / fill while a live view is wanted. Returns "
+            "path, bytes, url, title and timestamp."
+        ),
+        "inputSchema": {
+            "type": "object",
+            "properties": {
+                "output_filename": {"type": "string",
+                                    "description": "Plain basename "
+                                    "to save as. Default: "
+                                    "browser-live.png"},
+            },
+        },
+    },
     {
         "name": "browser_screenshot",
         "description": (
@@ -199,6 +237,11 @@ def handle_request(request, browser=None):
             elif name == "browser_fill":
                 result = target.fill(args.get("selector"),
                                      args.get("text"))
+            elif name == "browser_click_at":
+                result = target.click_at(args.get("x"), args.get("y"))
+            elif name == "browser_live_shot":
+                result = target.live_shot(
+                    args.get("output_filename"))
             elif name == "browser_evaluate":
                 result = target.evaluate(args.get("expression"))
             elif name == "browser_screenshot":