SkillAgentSearch skills...

droidpilot

Stable Android device automation for AI agents. Uses Accessibility Service + MCP protocol — no ADB, no OCR, just native OS APIs over WiFi.

Install / Use

claude mcp add youichi-uda -- npx -y github:youichi-uda/droidpilot

If the server publishes to npm under a different name, use that package instead — check the repo README.

About this skill
🔌

MCP Server

Model Context Protocol server

Quality Score

75/100

Category

Automation

Supported Platforms

Claude Code
Claude Desktop

Our assessment of droidpilot

droidpilot scores 75/100 on our quality scale, 2565th of 2,790 Automation skills we index.

Its MCP Server is 5.8 KB long, well organised into 16 sections with 9 code examples: a solid amount of guidance for an agent.

It has 10 GitHub stars, so there is little community track record yet; judge it on its content.

Substance
26/30
Structure
20/20
Description
15/15
Adoption
4/20
Freshness
11/15

Maintenance, license and trust

  • The repository was last updated about 5 months ago. That is recent enough to be usable, but agent tooling moves fast, so check the instructions against your agent's current version.
  • Our last check on 2026-11-11 found the source still online.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 95/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-11-10. Automated pattern scan on 2026-11-10. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

droidpilot compared with similar skills

All 4 of these similar skills score higher than droidpilot; compare them before choosing.

SkillScoreStarsUpdatedFormat
droidpilot (this skill)by youichi-uda75105mo agoMCP Server
Agent-Reachby Panniantong100112.3k8d agoCLAUDE.md
rufloby ruvnet10081.9ktodayCLAUDE.md
headroomby headroomlabs-ai10079.8ktodayCLAUDE.md
CowAgentby zhayujie10048.0ktodayCLAUDE.md

Frequently asked questions

How do I install droidpilot?
Run claude mcp add youichi-uda -- npx -y github:youichi-uda/droidpilot. The install tabs above show the steps for each supported agent.
Which AI agents does droidpilot work with?
It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
Is droidpilot safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 95/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is droidpilot still maintained?
The repository was last updated about 5 months ago. That is recent enough to be usable, but agent tooling moves fast, so check the instructions against your agent's current version.

DroidPilot

License: MIT Android 11+ MCP Compatible

Stable Android device automation for AI agents via Accessibility Service + MCP (Model Context Protocol).

Control any Android device from Claude, ChatGPT, or any MCP-compatible AI — no ADB, no USB, no screen mirroring. Just WiFi.

DroidPilot uses Android's native Accessibility Service to directly access the UI tree and perform gestures through OS APIs. This is fundamentally more reliable than ADB-based or OCR-based approaches used by other mobile automation tools.

Key Features

  • No ADB required — communicates over WiFi via WebSocket
  • Native UI tree access — no screenshot OCR or computer vision needed
  • Reliable gesture execution — taps, swipes, and text input via OS APIs
  • MCP native — works with Claude Desktop, Claude Code, and any MCP client
  • 18 automation tools — tap, swipe, type, screenshot, find element, and more
  • Low token cost — structured UI data instead of expensive image analysis
  • Simple setup — install APK, enable service, connect

How It Works

┌──────────────┐     MCP/stdio     ┌──────────────┐    WebSocket    ┌──────────────────┐
│  AI Agent    │ ◄──────────────► │  MCP Server  │ ◄────────────► │  Android Device  │
│  (Claude,    │                   │  (Node.js)   │    WiFi/LAN    │  (Accessibility   │
│   ChatGPT)   │                   │              │                │   Service + WS)   │
└──────────────┘                   └──────────────┘                └──────────────────┘

Why DroidPilot?

| Approach | Reliability | Speed | LLM Token Cost | Setup | |----------|-------------|-------|----------------|-------| | ADB-based (droidrun etc.) | Low — connection drops, limited UI access | Medium | High (screenshot analysis) | USB/WiFi ADB | | Screen mirroring + OCR | Low — OCR errors, high latency | Slow | Very High | Complex | | DroidPilot (Accessibility Service) | High — native OS integration | Fast | Low (structured data) | Install APK |

Available MCP Tools

| Tool | Description | |------|-------------| | connect | Connect to Android device by IP | | disconnect | Disconnect from device | | get_device_info | Device manufacturer, model, screen size, Android version | | screenshot | Capture screen as base64 JPEG image | | get_ui_tree | Full UI hierarchy with all element properties | | find_element | Search elements by text, ID, class, content description | | tap | Tap at screen coordinates | | long_press | Long press at coordinates | | swipe | Swipe gesture from point A to B | | scroll | Scroll in a direction (up/down/left/right) | | pinch | Pinch zoom in/out | | type_text | Append text to currently focused input | | set_text | Replace text in focused input | | press_key | System keys: back, home, recents, notifications, etc. | | click_element | Find and click element by text/ID (more reliable than coordinates) | | wait_for_element | Wait for element to appear on screen (with timeout) | | open_app | Launch app by package name | | get_focused | Get info about currently focused input element |

Quick Start

1. Android APK

Requirements: Android 11+ (API 30+), WiFi (same network as PC)

cd android
./gradlew assembleDebug
adb install app/build/outputs/apk/debug/app-debug.apk

Or open android/ in Android Studio and build from there.

Then on the device:

  1. Open DroidPilot app
  2. Tap "Open Accessibility Settings"
  3. Enable "Mobile MCP Pro"
  4. Return to app, tap "Start Server"
  5. Note the IP address displayed

2. MCP Server

cd mcp-server
npm install
npm run build

3. Configure Your AI Client

Claude Desktop — add to claude_desktop_config.json:

{
  "mcpServers": {
    "droidpilot": {
      "command": "node",
      "args": ["/path/to/droidpilot/mcp-server/dist/index.js"]
    }
  }
}

Claude Code — add to MCP settings:

{
  "mcpServers": {
    "droidpilot": {
      "command": "node",
      "args": ["/path/to/droidpilot/mcp-server/dist/index.js"]
    }
  }
}

4. Use It

Tell the AI:

Connect to my Android device at 192.168.1.100

Then give natural language commands:

Take a screenshot of the current screen
Open Chrome and navigate to google.com
Find the search bar and type "hello world"
Scroll down the page
Press the back button

Protocol

Communication between MCP Server and Android uses JSON over WebSocket:

Request:

{
  "id": "req_1_1234567890",
  "command": "tap",
  "params": { "x": 500, "y": 1000 }
}

Response:

{
  "id": "req_1_1234567890",
  "success": true,
  "data": { "action": "tap(500.0, 1000.0)" }
}

Use Cases

  • AI-powered mobile testing — let AI agents run QA flows on real devices
  • Mobile RPA — automate repetitive tasks across any Android app
  • Accessibility automation — build assistive workflows for users
  • App monitoring — periodic screenshots and UI state checks
  • Cross-app workflows — orchestrate actions across multiple apps

Security

  • WebSocket runs on local network only (no internet exposure)
  • Optional auth token support for WebSocket connections
  • No data is sent to external servers
  • All communication stays between your PC and your device on your LAN

Tech Stack

  • Android: Kotlin, AccessibilityService, Java-WebSocket
  • MCP Server: TypeScript, Node.js, @modelcontextprotocol/sdk
  • Communication: WebSocket (JSON protocol)

Contributing

Contributions are welcome! Feel free to open issues and pull requests.

License

MIT

Related Skills

View on GitHub
GitHub Stars10
CategoryAutomation
Updated5mo ago
Forks11

Languages

Kotlin

Trust signals

95/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

2 info