android-mcp-toolkit
A growing collection of MCP tools for Android Development. Currently features a deterministic Figma-SVG-to-Android-XML converter, with plans for Gradle analysis, Resource management, and ADB integration tools.
Android MCP Toolkit for AI Agents
Small MCP server with three tools:
- Fast SVG → Android VectorDrawable conversion (cached, file or inline).
- adb logcat reader with package/pid/tag filters for quick crash triage.
- Translation length difference estimator to flag risky length deltas before layout breaks.
Why this exists
The Mission: Bringing Native Android to the AI Agent Era
While the AI ecosystem flourishes with web-first tools, Android development often feels left behind. This MCP server is my answer to that gap—a dedicated bridge connecting AI Agents directly to the Android toolchain.
⚡ Zero-Friction Asset Conversion: Convert SVGs to VectorDrawables instantly without the overhead of launching Android Studio.
🔍 Direct Diagnostic Access: Empower agents to pull, filter, and analyze adb logcat streams (by package, PID, or tag) in real-time.
🤖 Agent-Native Architecture: Deliver structured, scriptable outputs that LLMs can parse and act upon efficiently.
🚀 Built for Extensibility: A solid foundation designed to grow, paving the way for future utilities like bitmap helpers and asset validation.
Pairing ideas
- Figma MCP: grab SVGs from designs, feed to
convert-svg-to-android-drawableto get XML for Android resources. - Debugging: while running the app, call
read-adb-logcatwith package name or tag to capture crashes and filtered logs without leaving the MCP client.
Previews
SVG to VectorDrawable
- Figma request → SVG extraction

- Flag conversion preview (single)

- Batch flag review (side-by-side)

- Batch run via MCP (console)

ADB logcat tool
- Crash capture prompt (inputs + filters)

- Response preview (summarized logcat)

Current tools
-
convert-svg-to-android-drawable- Inputs:
svg(inline) orsvgPath(file path). Optional:outputPath,floatPrecision(default 2),fillBlack(default false),xmlTag(default false),tint,cache(default true). - Output: VectorDrawable XML text; also writes to disk when
outputPathis provided. - Performance: LRU cache (32 entries) keyed by SVG + options plus fast reuse in-session.
- Converter: vendored fork in
vendor/svg2vectordrawablewith fixes forrgb()/rgba(),hsl()/hsla(), and named colors. Upstream license:vendor/svg2vectordrawable/LICENSE(MIT).
- Inputs:
-
manage-logcat- Inputs:
action:read(default),crash,anr, orclear.packageName: Optional. Resolves PID viaadb shell pidof.pid: Optional. Explicit PID.tag: Optional. Filter by tag (e.g.MyApp).priority:V,D,I,W,E,F,S(defaultV).maxLines: Tail count (default 200, max 2000).timeoutMs: Default 5000.
- Behavior:
read: Fetches logcat tail.crash: Fetcheslogcat -b crash.anr: Fetches recent ActivityManager ANR logs + tail of/data/anr/traces.txt.clear: clears logcat buffers.
- Inputs:
-
get-current-activity- Inputs:
timeoutMs(default5000, max15000). - Behavior: Inspects
dumpsys windowto find the currently focused app/window. Useful to verify state.
- Inputs:
-
dump-ui-hierarchy- Inputs:
timeoutMs(default 10000). - Behavior: Captures current UI hierarchy as XML via
uiautomator.
- Inputs:
-
take-screenshot- Inputs:
outputPath(required),timeoutMs. - Behavior: Saves device screenshot to local file.
- Inputs:
-
inject-input- Inputs:
command(tap,text,swipe,keyevent,back,home),args(array),timeoutMs. - Optional:
elementIdorelementText(finds element center and taps it). - Behavior: Simulates user interaction suitable for testing flows.
- Inputs:
-
estimate-text-length-difference- Inputs:
sourceText(original),translatedText(to compare),tolerancePercent(default30, max500). - Behavior: Measures grapheme length of both strings, computes percent change, and reports whether it exceeds the tolerance (useful to catch translation length blowups that could break layouts).
- Inputs:
Roadmap (planned)
- Additional MCP tools for Android assets (e.g., batch conversions, validations, optimizers).
- Optional resource prompts for common Android drawables/templates.
Quick start
npm installnpm run buildnode dist/index.js(stdio MCP server)
Run via npx
- Global:
npx android-mcp-toolkit
Use in Cursor (MCP config)
Add to your Cursor settings JSON:
{
"mcpServers": {
"android-mcp-toolkit": {
"command": "npx",
"args": [
"-y",
"android-mcp-toolkit"
]
}
}
}
The npx call downloads the published package; no local path required.
Quick install via Cursor deep link:
cursor://anysphere.cursor-deeplink/mcp/install?name=android-mcp-toolkit&config=eyJjb21tYW5kIjoibnB4IC15IGFuZHJvaWQtbWNwLXRvb2xraXQifQ%3D%3D
Examples
- Input SVG:
sample_svg.svg - Output VectorDrawable:
examples/sample_svg.xml
Notes
- Transport: stdio via
@modelcontextprotocol/sdk. - Base deps kept minimal; everything needed to convert SVGs is vendored/included.
Contact
Related Servers
Scout Monitoring MCP
sponsorPut performance and error data directly in the hands of your AI assistant.
Alpha Vantage MCP Server
sponsorAccess financial market data: realtime & historical stock, ETF, options, forex, crypto, commodities, fundamentals, technical indicators, & more
Just Prompt
A unified interface for various Large Language Model (LLM) providers, including OpenAI, Anthropic, Google Gemini, Groq, DeepSeek, and Ollama.
Petclinic
Interacts with the Swagger Petstore API using Petclinic v3 APIs, exposing tools for OpenAI models.
SuperCollider MCP Server
An MCP server for the SuperCollider programming language that executes synths using supercolliderjs.
Chronos Protocol
A robust MCP server that eliminates temporal blindness in AI coding agents through intelligent time tracking, persistent memory, and complete session traceability.
Netmind Code Interpreter
Execute code using the Netmind API.
Bifrost VSCode Dev Tools
Exposes VSCode dev tools features to MCP clients, with support for project-specific configurations.
Laravel Forge MCP Server
Manage Laravel Forge servers and sites using the Forge API.
AgentOps MCP
An MCP server for AgentOps, providing observability and evaluation tools for AI agents.
MCP QEMU VM Control
Give your AI full computer access — safely. Let Claude (or any MCP-compatible LLM) see your screen, move the mouse, type on the keyboard, and run commands — all inside an isolated QEMU virtual machine. Perfect for AI-driven automation, testing, and computer-use experiments without risking your host system.
Laburen MCP Server
A template for deploying a remote, authentication-free MCP server on Cloudflare Workers.