// SCAN REPORT

Claude Mobile

214 tools. 147 can modify or destroy data without limits.

20 destructive tools with no built-in limits. Policy required.

Last updated: 27 Sept 2026

147 can modify or destroy data

67 read-only

214 tools total

Community server · catalogue entry checked 23/09/2026

How to control Claude Mobile ↓

TOOL MAP

What Claude Mobile exposes to your agents

Read (67) Write / Execute (127) Destructive / Financial (20)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

Critical Risk

The most dangerous Claude Mobile tools

Destructive · High system_clear_logs Permanently removes device logs, destroying audit trails and forensic evidence irreversibly. Destructive · High app Uninstall action irreversibly removes apps; most severe applicable category applies. Destructive · High app_uninstall Irreversibly removes application and associated data from device without recovery option. Destructive · High browser_clear_session The tool permanently deletes stored browser session data without undo capability.

147 of Claude Mobile's 214 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Claude Mobile

PolicyLayer is an MCP gateway — it sits between your AI agents and Claude Mobile, and nothing reaches the server without passing your rules. These are the rules we recommend:

Deny destructive operations

{
  "app": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations

{
  "app_install": {
    "limits": [
      {
        "counter": "app_install_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "accessibility": {
    "limits": [
      {
        "counter": "accessibility_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register Claude Mobile — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON CLAUDE MOBILE →

Instant setup, no code required.

FULL CATALOGUE

All 214 Claude Mobile tools

DESTRUCTIVE 20 tools

Destructive app App lifecycle and inventory: launch, stop, install, uninstall, list Destructive app_uninstall Uninstall an app by package name or bundle ID Destructive browser_clear_session Delete all stored data for a session Destructive debug_remove_break Remove a breakpoint by its id (returned by debug_break). Destructive network_proxy Get, set, or clear the global HTTP proxy for the Android device. Destructive performance_heap_delete Delete a captured heap artifact and its metadata immediately. Destructive performance_trace_delete Delete a captured native performance trace artifact and its metadata. Destructive permission_reset Reset all permissions for an app Destructive permission_revoke Revoke app permission Destructive recorder Test scenario recorder. start: begin recording. stop: save scenario. status: check state. add_step: manual ste Destructive recorder_delete Delete a saved scenario Destructive recorder_remove_step Remove a step from the active recording by index Destructive repl_kill Terminate a REPL session (SIGTERM, then SIGKILL). Destructive rustore_discard Delete RuStore version draft Destructive sensor_battery Set battery level, charging status, and plugged state on Android. Changes persist until reset:true is used or Destructive store_discard Discard Google Play release draft Destructive sync_destroy Destroy a sync group and release resources. Destructive system_clear_logs Clear the device log buffer Destructive visual Visual regression testing. baseline_save: capture baseline screenshot. compare: diff against baseline. baselin Destructive visual_delete Delete a visual baseline

EXECUTE 86 tools

Execute app_launch Launch app by package name or bundle ID Execute app_restart Force-stop then re-launch an app. Common pattern for clearing in-memory state without uninstall. Execute app_stop Force stop an app Execute appstore_build Build a signed .ipa for TestFlight (auto-detects Flutter/React Native/KMP/Xcode projects). Execute appstore_submit Submit a TestFlight build for external beta review (defaults to the latest VALID build). Execute appstore_upload Validate then upload an .ipa to App Store Connect (TestFlight) via altool. Execute autopilot AI Test Autopilot. explore: auto-navigate app and build navigation graph. generate: create test scenarios from Execute autopilot_explore Automatically navigate the app, building a navigation graph of screens and transitions. Uses screen fingerprin Execute autopilot_heal Self-heal a broken test step by finding the best matching element on the current screen. Uses fuzzy matching o Execute browser Browser automation. open/close/list_sessions: session management. navigate/click/fill/fill_form/press_key: int Execute browser_click Click element by ref, selector, or text Execute browser_close Close browser session Execute browser_evaluate Execute JavaScript in browser page Execute browser_navigate Navigate to URL or go back/forward/reload Execute browser_open Open URL in browser session Execute browser_press_key Press keyboard key in browser Execute browser_wait_for_selector Wait for element to appear on page Execute clipboard_select Select all text in focused field (Android only) Execute debug_attach Attach the runtime debugger to a DEBUGGABLE running app. Android (JDWP): the package must have android:debugga Execute debug_break Set a breakpoint. Android: {className, line} or {className, method} (method-entry, robust without line info). Execute debug_detach Detach the debugger and end the session (the app keeps running). Also tears down the adb forward / LLDB daemon Execute debug_eval Evaluate an expression on a paused thread. Android: a local name, Execute debug_resume Resume all threads in the debugged VM/process. Use after inspecting a paused state to let the app continue run Execute debug_set_var Mutate a local variable on a paused thread. Android: primitives (int/long/bool/float/…), null, and strings; th Execute debug_step Step the paused thread: OVER (next line), INTO (into call), or OUT (out of current method). Issues the step an Execute desktop Desktop app management: launch (Gradle/bundle/attach), stop, windows, focus, resize, clipboard, performance, m Execute desktop_focus Focus a desktop window Execute desktop_launch Start desktop automation and optionally launch an app. Supports three modes: Execute desktop_resize Resize a desktop window Execute desktop_stop Stop running desktop application Execute device Device management + module loading. list/set/set_target/get_target: devices. enable_module/disable_module/list Execute flow Flow orchestration: batch (multi-command), run (multi-step automation with loops/conditionals), parallel (same Execute flow_batch Execute multiple commands in one round-trip. Set turbo:true for UI context per step (experimental). Execute flow_parallel Run same action on multiple devices in parallel. Uses Promise.allSettled for concurrent execution. Execute flow_run Multi-step automation flow with conditionals, loops, error handling. Use for E2E testing instead of calling to Execute go_back Press back button Execute harmony_launch_ability Launch a HarmonyOS ability with an optional module name. Defaults to EntryAbility. Execute harmony_test Run an ArkXTest module through Execute input Input actions. tap/double_tap/long_press: coords or text/id/label/index. swipe: direction or coords. text: typ Execute input_double_tap Double tap by coordinates, text, resourceId, or index. Raw x/y are screenshot-space and auto-scaled to device Execute input_key Press hardware key (BACK, HOME, ENTER, etc.) Execute input_long_press Long press at coordinates or on element by text/label. Raw x/y are screenshot-space and auto-scaled to device Execute input_swipe Swipe by direction or custom coordinates. Raw x1/y1/x2/y2 are screenshot-space and auto-scaled to device coord Execute input_tap Tap by coordinates, text, resourceId, label, or element index.\n\n Execute input_text Type text into focused input field Execute intent Intent & Deep Link Engine. start: launch activity with extras. broadcast: send broadcast intent. deeplink: ope Execute intent_broadcast Send an Android broadcast intent. Useful for triggering system events or communicating with broadcast receiver Execute intent_deeplink Open a deep link URI on Android or iOS. On Android uses Execute intent_start Launch an Activity with a structured Android Intent. Supports action, component, data URI, category, typed ext Execute launch_app Launch app Execute network Network Layer. traffic: app network traffic stats. connectivity: connection type/state. proxy: get/set HTTP pr Execute network_airplane Enable or disable airplane mode on the Android device. Execute performance Performance Lab. Metrics, baselines, monitoring, crashes, Android framestats, bounded native Perfetto/xctrace/ Execute performance_heap_capture Capture a private heap artifact without returning raw heap contents through MCP. Android requires a debuggable Execute performance_trace_start Start a bounded native performance trace. Android records Perfetto, iOS records an Instruments xctrace bundle, Execute performance_trace_stop Finalize an active performance trace, store the native artifact with mode 0600 and 24-hour TTL, and return a b Execute permission_grant Grant app permission (Android runtime / iOS privacy) Execute press_key Press key Execute recorder_play Replay a saved scenario. Executes all steps sequentially with optional speed/timeout control. Execute recorder_start Begin recording user interactions as a test scenario Execute repl_expect Block until a prompt regex matches, the session idles, the child exits, or the timeout fires. Execute repl_key Send a named key to a session. Editing/navigation for driving TUIs. Execute repl_resize Resize the PTY and vt100 grid for a live session. Execute repl_send Write text to a REPL session. Appends a newline by default. Execute repl_spawn Start an interactive REPL or CLI process under a PTY. Returns the session id. Execute sensor Sensor & Environment Simulation. location: set GPS coordinates. battery: set battery level/status. notificatio Execute sensor_location Set GPS location on device. Android emulator: uses Execute sensor_thermal Override Android thermal status (API 29+ / Android 10+). Simulates device overheating scenarios. Use reset:tru Execute store_halt_rollout Halt staged rollout on Google Play Execute swipe Swipe in a direction Execute sync Multi-device sync testing. create_group: bind devices to roles. run: coordinated steps with barriers. assert_c Execute sync_assert_cross Cross-device assertion: perform action on source device, verify result on target device with retries. Execute sync_run Execute coordinated steps across devices with barrier synchronization. Execute system System operations, clipboard, permissions, and file transfer. shell: run command. logs/wait_log: device logs. Execute system_open_url Open URL in device browser Execute system_shell Execute a shell command on the device. SECURITY: shell metacharacters Execute system_wait Wait for specified duration (ms) Execute tap Tap at x,y Execute tap_text Find and tap element by text Execute type_text Type into focused field Execute ui UI inspection and interaction. tree: accessibility tree. find: search elements. find_tap: fuzzy tap (Android/H Execute ui_find_tap Fuzzy tap by natural language element description (Android/HarmonyOS). When the matched element is a non-click Execute ui_tap_text Tap element by text via Accessibility API (Desktop/macOS only) Execute ui_wait Wait for UI element to appear (polling with timeout) Execute visual_suite Run visual comparison for all baselines matching filters (batch regression check) Execute wait Wait milliseconds

WRITE 41 tools

Write app_install Install APK (Android), .app bundle (iOS), RPM (Aurora), or HAP (HarmonyOS) Write appstore_promote Promote a TestFlight build: add it to a beta group by group name Write appstore_set_notes Set TestFlight Write autopilot_generate Generate test scenarios from exploration data. Creates flow_run-compatible test steps for all unique paths. Write browser_fill Fill input field with value Write browser_fill_form Fill multiple form fields at once Write clipboard_copy Select all and copy to clipboard (Android only) Write clipboard_paste Paste clipboard into focused field (Android only) Write clipboard_set Set clipboard text (Desktop only) Write debug_pause_state Inspect a paused thread: call stack frames (class, method, line) and the top frame Write device_set Select active device for subsequent commands. Sets global state — all following tool calls will target this de Write device_set_target Switch active platform (android/ios/desktop/aurora/harmony/browser) Write harmony_arkweb_close Remove an ArkWeb HDC port-forward created for inspection. Write harmony_sandbox_pull Download a file from a running debug-signed HarmonyOS application Write harmony_sandbox_push Upload a file into a running debug-signed HarmonyOS application Write huawei_set_notes Set release notes for Huawei AppGallery draft Write huawei_submit Submit Huawei AppGallery draft for review Write huawei_upload Upload APK/AAB to Huawei AppGallery. Requires HUAWEI_CLIENT_ID env. Write performance_baseline Save current performance metrics as a named baseline for later comparison. Write recorder_add_step Manually add a step to the active recording Write recorder_export Export scenario as flow_steps (for flow_run) or markdown checklist Write recorder_stop Stop recording and save scenario (or discard) Write rustore_set_notes Set what Write rustore_submit Submit RuStore draft for moderation Write rustore_upload Upload APK/AAB to RuStore. Requires RUSTORE_KEY_JSON env. Write sandbox App Sandbox Access. Write sandbox_prefs_write Write or update a single value in an app Write screen Screen capture and annotation. capture: screenshot with compression/diff. annotate: screenshot with element bo Write screen_capture Take screenshot. Auto-compressed. Use diff=true to see only changes. Write store App store management (Google Play, Huawei, RuStore, Apple TestFlight). Write store_promote Promote release between Google Play tracks Write store_set_notes Set release notes for Google Play draft (per language, max 500 chars) Write store_submit Publish release draft to Google Play track Write store_upload Upload APK/AAB to Google Play. Requires GOOGLE_PLAY_KEY_FILE env. Write sync_create_group Create a sync group of 2+ devices with named roles for coordinated testing. Write system_file_pull Download a file from a device with file transfer support Write system_file_push Upload a local file to a device with file transfer support Write ui_assert_gone Assert element does NOT exist on screen (pass/fail) Write ui_assert_visible Assert element is visible on screen (pass/fail) Write visual_baseline_save Save current screenshot as visual baseline for regression testing Write visual_baseline_update Update existing baseline with current screenshot

READ 67 tools

Read accessibility Accessibility Guardian. audit: full WCAG audit of current screen. check: audit a single element. summary: quic Read accessibility_audit Run full accessibility audit on current screen. Returns score, issues grouped by severity, and passed rules. Read accessibility_check Check accessibility of a specific element found by text, resourceId, or index. Read accessibility_rules List all accessibility rules or show details of a specific rule. Read accessibility_summary Quick accessibility summary: score + issue counts only (short output). Read app_list List installed apps on a platform with app inventory support Read appstore_get_releases List TestFlight builds with processing state. Read autopilot_status Get exploration status. If explorationId is provided, shows details. Otherwise lists all explorations. Read autopilot_tests List or get generated test scenarios for an exploration. Read browser_list_sessions List active browser sessions Read browser_screenshot Take browser page screenshot Read browser_snapshot Get accessibility snapshot with element refs Read clipboard_get Get clipboard text (Desktop only) Read clipboard_get_android Read clipboard text from Android device Read debug_poll Poll the event queue for hits (BREAKPOINT_HIT / STEP_HIT / EXCEPTION_HIT / CLASS_PREPARE / VM_DEATH). Never bl Read debug_sessions List active debug sessions (sessionId + platform). Useful to discover open sessions before attaching or after Read debug_threads List threads of the debugged VM with ids and names (Android). Use an id with debug_pause_state / debug_step to Read desktop_get_target_pid Get the PID of the native app set by the last desktop_launch (bundle or attach mode). Returns null if no targe Read desktop_monitors List connected monitors (Desktop only) Read desktop_performance Get memory and CPU metrics (Desktop only) Read desktop_windows Get desktop window info Read device_get_target Get current active platform and status Read device_info Get device info Read device_list List connected devices and emulators Read find_element Find element by text Read get_ui Get UI elements on screen Read harmony_arkweb_inspect Discover an ArkWeb DevTools socket, forward it over HDC, and list inspectable pages. Read harmony_sandbox_list List files in a running debug-signed HarmonyOS application Read harmony_sandbox_read Read a text file from a running debug-signed HarmonyOS application Read huawei_get_releases Get release info from Huawei AppGallery Read intent_services List running Android services. Optionally filter by package name. Android only. Read network_connectivity Get current network connectivity info: active network type (WiFi/Mobile/etc), connection state, IP address, DN Read network_traffic Get network traffic statistics. If a package name is provided, shows per-app traffic (rx/tx bytes and packets) Read performance_compare Compare current performance against a saved baseline. Returns PASS/FAIL per metric with thresholds. Read performance_crashes Query recent crashes, ANRs, and native crashes from device logs. Read performance_framestats Collect frame rendering statistics from GPU profiling. Returns frame time percentiles (p50/p90/p99), jank rate Read performance_heap_diff Compare bounded metadata from two compatible heap artifacts. Returns signed size/count deltas; it does not cla Read performance_monitor Monitor performance over a duration, collecting periodic samples. Returns min/max/avg stats. Read performance_snapshot Collect current performance metrics: memory, CPU, FPS, battery, crash count. Returns formatted report. Read performance_trace_status List active performance traces and the remaining capture window. Read recorder_list List saved test scenarios Read recorder_show Display contents of a saved scenario Read recorder_status Get current recording state Read repl_list List active REPL sessions and their statuses. Read rustore_get_versions Get version list from RuStore Read sandbox_file_list List files inside an app Read sandbox_file_read Read the contents of a file from an app Read sandbox_prefs_read Read SharedPreferences XML from an app Read sandbox_sqlite_query Run a read-only SQL query against an app Read screen_annotate Screenshot with numbered bounding boxes on UI elements (Android/iOS) Read screenshot Take screenshot Read sensor_notifications Read the notification shade from Android. Returns a parsed list of active notifications with title, text, pack Read store_get_releases Get current releases across Google Play tracks Read sync_list List all active sync groups. Read sync_status Show details of a sync group and its last run result. Read system_activity Get current foreground activity (Android only) Read system_info Get battery and memory info Read system_is_running Check whether an app process is currently running by package name (Android only). Returns Read system_logs Get device logs with optional filters Read system_pid_of Get the PID of a running app process by package name (Android only). Returns 0 when the package is not running Read system_wait_log Wait until a regex pattern appears in device logs. Polls the log buffer at regular intervals; returns the matc Read system_webview Inspect WebView via Chrome DevTools Protocol (Android only) Read ui_analyze Structured screen analysis: buttons, inputs, text, scrollable areas, dialogs Read ui_find Find UI elements by text, resourceId, className, or label Read ui_tree Get UI hierarchy (accessibility tree). Shows elements, text, IDs, coordinates. Read visual_compare Compare current screen with saved baseline. Returns diff overlay on mismatch. Read visual_list List saved visual baselines

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about Claude Mobile

Can an AI agent delete data through the Claude Mobile MCP server? +

Yes. The Claude Mobile server exposes 20 destructive tools including app, app_uninstall, browser_clear_session. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through Claude Mobile? +

The Claude Mobile server has 41 write tools including app_install, appstore_promote, appstore_set_notes. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach Claude Mobile.

How many tools does the Claude Mobile MCP server expose? +

214 tools across 4 categories: Destructive, Execute, Read, Write. 67 are read-only. 147 can modify, create, or delete data.

How do I enforce a policy on Claude Mobile? +

Register the Claude Mobile MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every Claude Mobile tool call.

Deterministic rules across all 214 Claude Mobile tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON CLAUDE MOBILE →

Instant setup, no code required.

214 Claude Mobile tools catalogued and risk-classified — across an index of 46,500+ MCP servers.