@nbakka/mcp-appium 2.0.10 → 2.0.12

package/lib/server.js

@@ -142,6 +142,22 @@ const createMcpServer = () => {
         const screenSize = await robot.getScreenSize();
         return `Screen size is ${screenSize.width}x${screenSize.height} pixels`;
     });
+tool(
+  "mobile_learn_current_app_context",
+  "Capture and store contextual knowledge about the current app screen or deeplink to inform intelligent automation decisions. This should be the first step after launching the app and before performing any tests.",
+  {},
+  async () => {
+    const context = {
+      notes: ["mira_road_east_buy_srp - https://housing.com/in/buy/mumbai/mira_road_east",
+        "Filters on the filter screen may initially be offscreen and require scrolling to view.",
+        "This context enables dynamic decisions during automation without hardcoded actions.",
+      ],
+    };
+
+    return `App context learned`;
+  }
+);
+
 
 tool("mobile_list_elements_on_screen", "List elements on screen and their coordinates, with display text or accessibility label. Do not cache this result.", {}, async ({}) => {
         requireRobot();
@@ -214,73 +230,34 @@ await new Promise(resolve => setTimeout(resolve, 5000));
         return `Current device orientation is ${orientation}`;
     });
 tool(
-  "tap_by_text",
-  "tap an element by passing the text visible on screen",
-  { text: zod_1.z.string().describe("The exact text to search for in the UI element"), },
+  "mobile_tap_by_text",
+  "Tap an element on screen by its displayed text or accessibility label using ADB tap",
+  {
+    text: zod_1.z.string().describe("The exact text or label of the element to tap"),
+  },
   async ({ text }) => {
-    const { execSync } = require("child_process");
-    const { XMLParser } = require("fast-xml-parser");
-
     if (!text) throw new Error("Input text is required");
 
-    // 1. Dump UI Automator XML to stdout
-    let dump;
-    try {
-      dump = execSync("adb shell uiautomator dump /dev/tty", { maxBuffer: 10 * 1024 * 1024 }).toString();
-    } catch (e) {
-      throw new Error("Failed to dump UI Automator XML via adb");
-    }
-
-    // 2. Parse XML
-    const parser = new XMLParser({
-      ignoreAttributes: false,
-      attributeNamePrefix: "",
-    });
-    const xmlObj = parser.parse(dump);
-
-    // 3. Recursive traversal to find matching element by text/content-desc/hint
-    function findElement(node) {
-      if (!node) return null;
-
-      const matchText = node.text || node["content-desc"] || node.hint || "";
-      if (matchText === text) return node;
+    requireRobot(); // ensure robot instance available
+    const elements = await robot.getElementsOnScreen();
 
-      if (node.node) {
-        if (Array.isArray(node.node)) {
-          for (const child of node.node) {
-            const found = findElement(child);
-            if (found) return found;
-          }
-        } else {
-          return findElement(node.node);
-        }
-      }
-
-      return null;
-    }
+    // Find element by exact match on text or label
+    const element = elements.find(
+      el => el.text === text || el.label === text
+    );
 
-    const element = findElement(xmlObj.hierarchy.node);
-    if (!element) throw new Error(`Element with text "${text}" not found`);
+    if (!element) throw new Error(`Element with text or label "${text}" not found`);
 
-    // 4. Parse bounds: format "[left,top][right,bottom]"
-    const bounds = element.bounds;
-    if (!bounds) throw new Error("Element bounds not found");
+    // Calculate center coordinates
+    const rect = element.rect;
+    const x = Math.floor(rect.x + rect.width / 2);
+    const y = Math.floor(rect.y + rect.height / 2);
 
-    const coords = bounds.match(/\d+/g).map(Number);
-    if (coords.length !== 4) throw new Error("Invalid bounds format");
-
-    const [left, top, right, bottom] = coords;
-    const x = Math.floor((left + right) / 2);
-    const y = Math.floor((top + bottom) / 2);
-
-    // 5. Tap via adb
-    try {
-      execSync(`adb shell input tap ${x} ${y}`);
-    } catch (e) {
-      throw new Error("Failed to perform tap via adb");
-    }
+    // Execute adb tap
+    const { execSync } = require("child_process");
+    execSync(`adb shell input tap ${x} ${y}`);
 
-    return `Tapped element with text "${text}" at (${x},${y})`;
+    return `Tapped element with text/label "${text}" at (${x},${y})`;
   }
 );
     // async check for latest agent version

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nbakka/mcp-appium",
-  "version": "2.0.10",
+  "version": "2.0.12",
   "description": "Appium MCP",
   "engines": {
     "node": ">=18"