Features · Installation · Tools · Usage · Contributing · Security
mobile-security-mcp is an MCP (Model Context Protocol) server that gives Claude — and any MCP-compatible AI client — the ability to analyze Android APK and iOS IPA files for security issues through natural language conversation.
Security researchers, mobile pentesters, and app developers can now audit permissions, extract API endpoints, detect hardcoded secrets, inspect Firebase configuration, and enumerate third-party SDKs by simply asking Claude — no scripting required.
| Tool | What it does |
|---|---|
apk_manifest_analyzer |
Parses AndroidManifest.xml — flags debuggable, allowBackup, exported components, intent filters |
apk_permissions_checker |
Categorizes all permissions into dangerous vs normal with risk explanations |
android_api_extractor |
Decompiles smali bytecode to extract Retrofit HTTP endpoints and OkHttp3 fields |
android_google_services |
Extracts Firebase/GCP config from google-services.json and resources.arsc string values |
android_secrets_scanner |
Scans DEX bytecode + resources.arsc + assets for hardcoded API keys and credentials |
| Tool | What it does |
|---|---|
ios_manifest_analyzer |
Parses Info.plist — flags ATS misconfigs, URL schemes, background modes |
ios_permissions_checker |
Categorizes privacy permission declarations by HIGH / MEDIUM / LOW risk |
ios_entitlements_checker |
Extracts entitlements via codesign — flags get-task-allow, sandbox bypass, iCloud containers |
ios_binary_strings |
Extracts URLs, emails, IPs, and API key patterns from the Mach-O binary |
ios_frameworks_detector |
Lists bundled frameworks, maps ~60 known SDKs (analytics, ads, attribution, crash reporting) |
ios_google_services |
Parses GoogleService-Info.plist for full Firebase configuration |
ios_secrets_scanner |
Scans app binary + resource files for hardcoded secrets and credentials |
All secret and Google service detection patterns live in a single patterns.ts — easy to extend, used by both Android and iOS scanners.
npm install -g mobile-security-mcpAdd to your claude_desktop_config.json:
{
"mcpServers": {
"mobile-security-mcp": {
"command": "npx",
"args": ["mobile-security-mcp"]
}
}
}Config file locations:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
git clone https://github.com/Serhatcck/mobile-security-mcp.git
cd mobile-security-mcp
npm install && npm run build{
"mcpServers": {
"mobile-security-mcp": {
"command": "node",
"args": ["/absolute/path/to/mobile-security-mcp/dist/index.js"]
}
}
}Once configured, restart Claude Desktop and start a conversation:
"Analyze the permissions in /path/to/app.apk"
"Check this IPA for hardcoded API keys: /path/to/app.ipa"
"What Firebase services does this APK use?"
"Are there any exported components in this APK that could be an attack surface?"
"Show me all third-party SDKs in this iOS app and flag any privacy risks"
Android:
apktool— required forandroid_api_extractor(brew install apktool)aapt(optional) — speeds up manifest parsing, part of Android SDK build tools
iOS (macOS only):
codesign,plutil,strings— all built into macOS, no install needed
Input: apk_path (string)
Output: Package info, security flags, components, intent filters, warnings
Input: apk_path (string)
Output: Dangerous permissions (highlighted) + normal permissions + risk summary
Input: apk_path OR smali_folder (string), output_format (txt|postman)
Output: Retrofit HTTP endpoints or Postman collection JSON
Input: apk_path (string), smali_folder (optional)
Output: Firebase project ID, API keys, database URL, storage bucket, OAuth clients
Input: apk_path (string), smali_folder (optional), min_length (default 8)
Output: Hardcoded credentials found in DEX + resources.arsc + assets
Input: ipa_path (string)
Output: Bundle info, ATS settings, URL schemes, background modes, warnings
Input: ipa_path (string)
Output: Privacy permissions grouped by HIGH/MEDIUM/LOW risk with usage descriptions
Input: ipa_path (string)
Output: Entitlements extracted from binary, high-risk flags, simulator detection
Input: ipa_path (string), filter (all|url|key|email|ip), min_length (default 6)
Output: Filtered strings from Mach-O binary
Input: ipa_path (string)
Output: Bundled frameworks grouped by category with privacy risk annotations
Input: ipa_path (string)
Output: Full GoogleService-Info.plist contents + pattern scan of resource files
Input: ipa_path (string), min_length (default 8)
Output: Secrets found in resource files and binary, split by layer with severity
Regenerate with VHS:
brew install charmbracelet/tap/vhs && vhs docs/demo.tape
See CONTRIBUTING.md for development setup, how to add new tools, and PR guidelines.
See SECURITY.md for how to report vulnerabilities privately.
See CHANGELOG.md.
