mobai-mcp 1.2.1 → 1.4.0

package/dist/index.js

@@ -238,6 +238,136 @@ const TOOLS = [
             required: ["device_id"],
         },
     },
+    {
+        name: "double_tap",
+        description: "Double tap an element by index (from UI tree) or coordinates",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+                index: {
+                    type: "number",
+                    description: "Element index from UI tree (preferred)",
+                },
+                x: {
+                    type: "number",
+                    description: "X coordinate (use with y instead of index)",
+                },
+                y: {
+                    type: "number",
+                    description: "Y coordinate (use with x instead of index)",
+                },
+            },
+            required: ["device_id"],
+        },
+    },
+    {
+        name: "long_press",
+        description: "Long press an element by index (from UI tree) or coordinates. Uses a fixed 0.5s hold duration.",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+                index: {
+                    type: "number",
+                    description: "Element index from UI tree (preferred)",
+                },
+                x: {
+                    type: "number",
+                    description: "X coordinate (use with y instead of index)",
+                },
+                y: {
+                    type: "number",
+                    description: "Y coordinate (use with x instead of index)",
+                },
+            },
+            required: ["device_id"],
+        },
+    },
+    {
+        name: "two_finger_tap",
+        description: "Perform a two-finger tap at coordinates (iOS only)",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+                index: {
+                    type: "number",
+                    description: "Element index from UI tree (preferred)",
+                },
+                x: {
+                    type: "number",
+                    description: "X coordinate (use with y instead of index)",
+                },
+                y: {
+                    type: "number",
+                    description: "Y coordinate (use with x instead of index)",
+                },
+            },
+            required: ["device_id"],
+        },
+    },
+    {
+        name: "drag",
+        description: "Drag from one point to another (press, hold, move, release)",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+                from_x: {
+                    type: "number",
+                    description: "Starting X coordinate",
+                },
+                from_y: {
+                    type: "number",
+                    description: "Starting Y coordinate",
+                },
+                to_x: {
+                    type: "number",
+                    description: "Ending X coordinate",
+                },
+                to_y: {
+                    type: "number",
+                    description: "Ending Y coordinate",
+                },
+                duration_ms: {
+                    type: "number",
+                    description: "Drag duration in milliseconds (default: 500)",
+                },
+                press_duration_ms: {
+                    type: "number",
+                    description: "Hold duration before dragging in milliseconds (0 = no hold). Use for press-and-drag gestures like moving app icons.",
+                },
+            },
+            required: ["device_id", "from_x", "from_y", "to_x", "to_y"],
+        },
+    },
+    {
+        name: "dismiss_keyboard",
+        description: "Dismiss the on-screen keyboard if visible",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+            },
+            required: ["device_id"],
+        },
+    },
     {
         name: "type_text",
         description: "Type text on the device (tap input field first to focus)",
@@ -355,7 +485,7 @@ const TOOLS = [
         description: `Execute a batch of automation steps using the DSL (Domain Specific Language).
 This is the PREFERRED method for complex automation as it's more reliable than sequential API calls.
 
-DSL supports: observe, tap, type, toggle, swipe, scroll, open_app, navigate, wait_for, assert_*, if_exists, delay, execute_js (web)
+DSL supports: observe, tap, type, toggle, swipe, scroll, open_app, kill_app, navigate, wait_for, screenshot, set_location, reset_location, assert_*, if_exists, delay, execute_js (web)
 
 Example DSL script:
 {
@@ -530,6 +660,78 @@ Example DSL script:
             required: ["device_id", "script"],
         },
     },
+    {
+        name: "uninstall_app",
+        description: "Uninstall an application from the device by bundle ID / package name.",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+                bundle_id: {
+                    type: "string",
+                    description: "App bundle ID (iOS) or package name (Android) to uninstall",
+                },
+            },
+            required: ["device_id", "bundle_id"],
+        },
+    },
+    {
+        name: "kill_app",
+        description: "Force-kill a running application. On iOS (17+), uses CoreDevice appservice SIGKILL. On Android, uses 'am force-stop'.",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+                bundle_id: {
+                    type: "string",
+                    description: "Bundle ID / package name of the app to kill",
+                },
+            },
+            required: ["device_id", "bundle_id"],
+        },
+    },
+    {
+        name: "set_location",
+        description: "Set a simulated GPS location on the device. Supports: iOS (all versions), Android emulators (all versions), Android real devices (12+ only).",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+                lat: {
+                    type: "number",
+                    description: "Latitude (-90 to 90)",
+                },
+                lon: {
+                    type: "number",
+                    description: "Longitude (-180 to 180)",
+                },
+            },
+            required: ["device_id", "lat", "lon"],
+        },
+    },
+    {
+        name: "reset_location",
+        description: "Reset the device location to its real GPS position, removing any simulated location. Supports: iOS (all versions), Android emulators (all versions), Android real devices (12+ only).",
+        inputSchema: {
+            type: "object",
+            properties: {
+                device_id: {
+                    type: "string",
+                    description: "Device ID",
+                },
+            },
+            required: ["device_id"],
+        },
+    },
     {
         name: "http_request",
         description: `Make a raw HTTP request to the MobAI API. Use this for advanced operations not covered by other tools.
@@ -616,6 +818,56 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
                 result = await makeRequest("POST", `/devices/${args?.device_id}/tap`, body);
                 break;
             }
+            case "double_tap": {
+                const body = {};
+                if (args?.index !== undefined)
+                    body.index = args.index;
+                if (args?.x !== undefined && args?.y !== undefined) {
+                    body.x = args.x;
+                    body.y = args.y;
+                }
+                result = await makeRequest("POST", `/devices/${args?.device_id}/double-tap`, body);
+                break;
+            }
+            case "long_press": {
+                const body = {};
+                if (args?.index !== undefined)
+                    body.index = args.index;
+                if (args?.x !== undefined && args?.y !== undefined) {
+                    body.x = args.x;
+                    body.y = args.y;
+                }
+                result = await makeRequest("POST", `/devices/${args?.device_id}/long-press`, body);
+                break;
+            }
+            case "two_finger_tap": {
+                const body = {};
+                if (args?.index !== undefined)
+                    body.index = args.index;
+                if (args?.x !== undefined && args?.y !== undefined) {
+                    body.x = args.x;
+                    body.y = args.y;
+                }
+                result = await makeRequest("POST", `/devices/${args?.device_id}/two-finger-tap`, body);
+                break;
+            }
+            case "drag": {
+                const dragBody = {
+                    fromX: args?.from_x,
+                    fromY: args?.from_y,
+                    toX: args?.to_x,
+                    toY: args?.to_y,
+                    duration: args?.duration_ms ?? 500,
+                };
+                if (args?.press_duration_ms) {
+                    dragBody.pressDuration = args.press_duration_ms;
+                }
+                result = await makeRequest("POST", `/devices/${args?.device_id}/drag`, dragBody);
+                break;
+            }
+            case "dismiss_keyboard":
+                result = await makeRequest("POST", `/devices/${args?.device_id}/dismiss-keyboard`);
+                break;
             case "type_text":
                 result = await makeRequest("POST", `/devices/${args?.device_id}/type`, { text: args?.text });
                 break;
@@ -642,6 +894,23 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
             case "get_ocr":
                 result = await makeRequest("GET", `/devices/${args?.device_id}/ocr`);
                 break;
+            case "uninstall_app":
+                result = await makeRequest("DELETE", `/devices/${args?.device_id}/apps/${encodeURIComponent(args?.bundle_id)}`);
+                break;
+            case "kill_app":
+                result = await makeRequest("POST", `/devices/${args?.device_id}/kill-app`, {
+                    bundleId: args?.bundle_id,
+                });
+                break;
+            case "set_location":
+                result = await makeRequest("POST", `/devices/${args?.device_id}/location`, {
+                    lat: args?.lat,
+                    lon: args?.lon,
+                });
+                break;
+            case "reset_location":
+                result = await makeRequest("DELETE", `/devices/${args?.device_id}/location`);
+                break;
             case "execute_dsl":
                 result = await makeRequest("POST", `/devices/${args?.device_id}/dsl/execute`, args?.script, 300000 // 5 minutes
                 );
@@ -822,10 +1091,17 @@ const API_REFERENCE = `# MobAI API Reference
 | Endpoint | Method | Description |
 |----------|--------|-------------|
 | /devices/{id}/tap | POST | Tap element: {"index": N} or {"x": X, "y": Y} |
+| /devices/{id}/double-tap | POST | Double tap: {"index": N} or {"x": X, "y": Y} |
+| /devices/{id}/long-press | POST | Long press (0.5s): {"index": N} or {"x": X, "y": Y} |
+| /devices/{id}/two-finger-tap | POST | Two-finger tap (iOS): {"index": N} or {"x": X, "y": Y} |
 | /devices/{id}/swipe | POST | Swipe: {"fromX", "fromY", "toX", "toY", "duration"} |
+| /devices/{id}/drag | POST | Drag: {"fromX", "fromY", "toX", "toY", "duration", "pressDuration"} |
 | /devices/{id}/type | POST | Type text: {"text": "..."} |
+| /devices/{id}/dismiss-keyboard | POST | Dismiss on-screen keyboard |
 | /devices/{id}/go-home | POST | Go to home screen |
 | /devices/{id}/launch-app | POST | Launch app: {"bundleId": "..."} |
+| /devices/{id}/apps/{bundleId} | DELETE | Uninstall app by bundle ID |
+| /devices/{id}/kill-app | POST | Kill app: {"bundleId": "..."} |
 
 ## DSL Execution
 
@@ -839,6 +1115,15 @@ const API_REFERENCE = `# MobAI API Reference
 |----------|--------|-------------|
 | /devices/{id}/agent/run | POST | Run AI agent: {"task": "..."} |
 
+## Performance Metrics
+
+| Endpoint | Method | Description |
+|----------|--------|-------------|
+| /devices/{id}/metrics/start | POST | Start metrics collection |
+| /devices/{id}/metrics/stop | POST | Stop collection, return summary |
+| /devices/{id}/metrics | GET | Get raw metrics buffer |
+| /devices/{id}/metrics/summary | GET | Get current summary without stopping |
+
 ## Web Automation
 
 | Endpoint | Method | Description |
@@ -922,12 +1207,18 @@ The DSL (Domain Specific Language) enables batch execution of multiple automatio
 | scroll | Scroll in container | direction, predicate (container), to_element |
 | open_app | Launch app | bundle_id |
 | navigate | Go home/back | target ("home", "back") |
-| wait_for | Wait for element | predicate, timeout_ms |
+| wait_for | Wait for element or UI stability | predicate, timeout_ms, poll_interval_ms, stable (wait for UI to stop changing) |
+| screenshot | Save screenshot to file | file_path (directory), name (optional filename) |
 | assert_exists | Verify element exists | predicate, timeout_ms |
 | assert_not_exists | Verify element gone | predicate |
 | delay | Wait fixed time | duration_ms |
 | if_exists | Conditional | predicate, then, else |
 | select_web_context | Select browser/WebView | url_contains, title_contains (optional filters) |
+| kill_app | Force-kill running app | bundle_id |
+| set_location | Simulate GPS location (Android 12+ for real devices) | lat, lon |
+| reset_location | Reset to real GPS (Android 12+ for real devices) | (no fields) |
+| metrics_start | Start performance monitoring | types, bundle_id, label, thresholds, capture_logs |
+| metrics_stop | Stop monitoring, get summary | format ("summary" or "detailed") |
 
 ## Predicates
 
@@ -990,6 +1281,86 @@ Use \`include: ["ocr"]\` in observe to get text recognition when UI tree is empt
 \`\`\`
 
 Returns text with coordinates for tapping (already adjusted for tapping).
+
+## Performance Metrics
+
+Collect CPU, memory, FPS, network, and battery metrics during test flows with optional logging capture.
+
+### Start Metrics Collection
+\`\`\`json
+{
+  "action": "metrics_start",
+  "types": ["system_cpu", "system_memory", "fps"],
+  "bundle_id": "com.example.app",
+  "label": "login_flow",
+  "capture_logs": true,
+  "thresholds": {
+    "cpu_high": 80,
+    "fps_low": 45,
+    "memory_growth_mb_min": 50
+  }
+}
+\`\`\`
+
+**Fields:**
+- \`types\`: Metrics to collect - system_cpu, system_memory, fps, network, battery, process
+- \`bundle_id\`: Filter to specific app (optional)
+- \`label\`: Human-readable session label (optional)
+- \`thresholds\`: Custom thresholds for anomaly detection (optional)
+- \`capture_logs\`: Capture device logs during session (default: false)
+
+### Stop and Get Summary
+\`\`\`json
+{"action": "metrics_stop", "format": "summary"}
+\`\`\`
+
+**Response:**
+\`\`\`json
+{
+  "metrics_summary": {
+    "session": {
+      "label": "login_flow",
+      "duration_seconds": 45.2,
+      "sample_count": 45,
+      "session_id": "abc123",
+      "data_file": "/tmp/mobai/metrics/abc123.jsonl",
+      "logs_file": "/tmp/mobai/logs/abc123.jsonl",
+      "logs_available": true
+    },
+    "overall_health": "warning",
+    "health_score": 72,
+    "system_cpu": {"avg": 34.5, "max": 89.2, "p95": 78.1, "status": "ok"},
+    "system_memory": {"avg_percent": 45.2, "growth_mb": 28.5, "trend": "increasing", "status": "warning"},
+    "fps": {"avg": 58.2, "min": 24.0, "jank_percent": 8.5, "status": "warning"},
+    "anomalies": {
+      "cpu_spikes": [
+        {"at_s": 0.5, "peak": 288, "duration_ms": 18147, "source": "system"}   
+      ],
+      "fps_drops": [
+        {"start_s": 1.2, "end_s": 16.8, "min_fps": 39.5, "avg_fps": 42.3, "samples": 1}
+      ],
+    },
+    "recommendations": [
+      "FPS dropped to 24 at +15s - investigate screen transition"
+    ]
+  }
+}
+\`\`\`
+
+### Example: Performance Test Flow
+\`\`\`json
+{
+  "version": "0.2",
+  "steps": [
+    {"action": "metrics_start", "types": ["system_cpu", "system_memory", "fps"], "label": "app_launch"},
+    {"action": "open_app", "bundle_id": "com.example.app"},
+    {"action": "wait_for", "predicate": {"text": "Welcome"}, "timeout_ms": 10000},
+    {"action": "tap", "predicate": {"text": "Login"}},
+    {"action": "delay", "duration_ms": 5000},
+    {"action": "metrics_stop", "format": "summary"}
+  ]
+}
+\`\`\`
 `;
 const NATIVE_RUNNER_GUIDE = `# Native App Automation Guide
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mobai-mcp",
-  "version": "1.2.1",
+  "version": "1.4.0",
   "mcpName": "io.github.MobAI-App/mobai-mcp",
   "description": "MCP server for MobAI - AI-powered mobile device automation",
   "type": "module",