LuckyStar MCP Server
MCP server that lets AI agents drive the LuckyStar/LS_KTool Android memory-debugging stack, exposing tools for memory read/write, scanning, pointer operations, and device/process management via the Model Context Protocol.
README
lucky-star-mcp
AI-agent plugins for the LuckyStar / LS_KTool Android memory-debugging stack (from lsnbm/Linux-android-arm64).
This repository packages the agent-facing components so any MCP-capable client (Claude Code, Claude Desktop, Cursor, etc.) can drive the Android debugger through the standard Model Context Protocol (MCP).
⚠️ Research/learning use only. The upstream driver provides kernel memory read/write, module scanning, hardware breakpoints, and input simulation. Use it only on devices and processes you own or are authorized to test.
What's inside
lucky-star-mcp/
├── python/
│ ├── LuckyStarMcp.py # FastMCP server → exposes the Android bridge as MCP tools
│ ├── http_bridge.py # NativeHttpBridge → talks to the Android HTTP server (port 9494)
│ └── examples/
│ └── mcp-config.example.json # sample MCP client registration
├── go-client/ # Go desktop client (Fyne UI) with a bundled MCP server (internal/mcp)
├── requirements.txt # Python deps (mcp, pydantic)
└── LICENSE # MIT
Python MCP server (python/)
LuckyStarMcp.py is a self-contained FastMCP server. It exposes the Android
debugger's full toolset as MCP tools:
- Connection:
android_connection,discover_android_bridges,android_bridge_ping - Target:
android_target_set_pid,android_target_attach_package,android_target_find_pid,android_target_current - Environment:
android_env_get_params - Memory:
android_memory_regions,android_memory_read,android_memory_write,android_memory_dump - Scanning:
android_memory_scan_start,android_memory_scan_refine,android_memory_scan_results,android_memory_scan_status,android_memory_scan_clear - Pointers:
android_pointer_status,android_pointer_scan,android_pointer_merge,android_saved_* - Modules/breakpoints:
android_module_address, breakpoint helpers
Go client (go-client/)
A desktop GUI client (Fyne) for the same bridge, with its own MCP server in
internal/mcp (uses mark3labs/mcp-go). Build with build.ps1 / go build.
Prerequisites
- Android device running
LS_KToolin HTTP-server mode (port9494), reachable over the LAN (or via a Cloudflare tunnel URL). - Python 3.12+ for the Python MCP server.
The MCP server starts even when the device is offline — each tool call will then return a bridge-connection error instead.
One-shot install for local agents
./install.sh sets up a Python venv, registers the MCP server with your local
agents (Claude Code, Claude Desktop, Cursor), and starts it:
./install.sh # venv + deps + register agents + start server
./install.sh status # check server / adb forward state
./install.sh stop # stop the server
It also runs adb forward tcp:9494 tcp:9494 so the server reaches the device
through the adb connection (--android-host 127.0.0.1).
Install & run (Python)
pip install -r requirements.txt
# Start the MCP server (default: http://127.0.0.1:14447/mcp)
python python/LuckyStarMcp.py
# Point it at a specific device on your LAN:
python python/LuckyStarMcp.py --android-host 192.168.1.50 --android-timeout 4
# Use a Cloudflare tunnel URL instead of a raw IP:
python python/LuckyStarMcp.py --android-host "https://your-tunnel.trycloudflare.com"
| Flag | Default | Description |
|---|---|---|
--mcp-host |
127.0.0.1 |
Bind host for the local MCP web server |
--mcp-port |
14447 |
Bind port for the local MCP web server |
--mcp-path |
/mcp |
HTTP endpoint path for streamable-http clients |
--android-host |
auto |
Android IP/host, full HTTP(S) tunnel URL, or auto for LAN discovery |
--android-timeout |
4 |
Timeout (seconds) for Android bridge requests |
Environment variables: ANDROID_HTTP_HOST (default auto), ANDROID_HTTP_PORT
(9494), ANDROID_HTTP_TIMEOUT (4), ANDROID_MCP_BIND_HOST (127.0.0.1),
ANDROID_MCP_BIND_PORT (14447), ANDROID_MCP_PATH (/mcp).
Register with an MCP client
With the server running, register it in your client's MCP config. Example (Claude Code / Claude Desktop style):
{
"mcpServers": {
"luckystar": {
"type": "http",
"url": "http://127.0.0.1:14447/mcp",
"enabled": true
}
}
}
Or, to have the client launch the server itself (stdio mode):
{
"mcpServers": {
"luckystar": {
"command": "python",
"args": ["/path/to/lucky-star-mcp/python/LuckyStarMcp.py"]
}
}
}
License
MIT. The LuckyStarMcp.py, http_bridge.py, and go-client/ sources are
derived from lsnbm/Linux-android-arm64
(© 2025 LsLucky). See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.