Android APK Risk Rules: Manifest and DEX References

    1

    A real stalkerware sample once scored 36/100 because the engine read only the bytecode and never the manifest.

    $39

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Android APK Risk Rules: Manifest and DEX References

    Example session with this skill installed

    Analyze this sideloaded APK file for potential surveillance capabilities and hidden API calls.

    • Read your context and instructions
    • Compiled the android apk risk

    The APK analysis is complete. I scanned the binary manifest and two DEX files, identifying a high-risk combination of data access permissions and corroborating method references in the bytecode.

    APK Risk Report

    MetricValue
    VerdictDANGEROUS
    Risk Score88/100
    Permissions Found14 (6 sensitive categories)
    DEX Files2 (215,400 bytes)
    Native Libraries2 (Not analyzed)

    Findings

    [Critical] APK-PERM-SURVEILLANCE-001

    • Description: APK declares broad personal-data access across 6 categories including Location, SMS, and Contacts.
    • Advice: Verify if the app’s primary function (e.g., a simple calculator) justifies this level of data access.

    [High] APK-DEX-GATED-SMS-004

    Description: Found android.telephony.SmsManager method references in DEX combined with RECEIVE_SMS permission.

    • Advice: High risk of intercepting two-factor authentication codes or private messages.

    [Medium] APK-DEX-MARKER-RUNTIME-002

    • Description: Detected DexClassLoader and System.loadLibrary calls.
    • Advice: The app can load and execute code at runtime that was not present during this scan.

    Coverage & Caveats

    • Coverage: 100% of manifest strings and DEX method indexes analyzed.

    Caveats: 2 native libraries (.so files) were detected but not analyzed. The manifest parser did not map permissions to specific activities or services.

    Next steps

    • Review the specific DEX method calls related to SMS handling in a full decompiler.
    • Check the native directory of the APK to identify the purpose of the unread .so files.
    • Cross-reference the behavior IDs with your internal malware database.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    A real stalkerware sample once scored 36/100 because the engine read only the bytecode and never the manifest. Permissions are the highest-quality signal an APK offers and the hardest to hide: obfuscation, reflection and native payloads defeat a string sweep, none of them defeat the manifest. This kit is the rule set built on that lesson, the scanner that feeds it and a report that is honest about its limits.

    apk-rules.mjs: manifest rules that count personal-data channels (location, microphone, camera, SMS, call log, contacts, calendar; four or more critical, three high) and flag accessibility, device admin, notification access, overlay only with accessibility, all-files storage, battery exemption only with sensitive access, package and usage visibility. DEX rules that need an API reference plus a marker plus, where Android gates the call, the permission: location, microphone and SMS code with their declarations; shell execution plus a superuser binary; debugger check plus emulator properties; component toggling (icon hiding); Dalvik class-loader constructors (runtime code loading); a cleartext endpoint with network code and INTERNET. Every finding carries what the user should do, never 'investigate further'. riskScore is driven by the worst finding with diminishing extras, so ten informational notes never dilute one critical; verdicts dangerous, concerning, notable, clean. mergeFindings keeps one row per behaviour.

    apk-evidence.mjs: a dependency-free reader for the APK zip (central directory, stored and deflated entries), the binary AndroidManifest.xml string pool (UTF-8 and UTF-16) for declared permissions, and the string, type and method indexes of every classes.dex with bounds on entries, bytes, characters and index items. Native libraries are counted, not read, and the report says so.

    scan-apk.mjs: CLI and JSON report with permissions, findings, markers, score, coverage (DEX count and bytes, truncation, unparsed files, native libraries, manifest found) and caveats. rule-catalog.json: the topic taxonomy, the 17 ids this kit emits and 37 reserved for a decompiling tier, under one contract: an id names a behaviour, not an engine.

    Reference: why the manifest comes first, why rules are combinations, why the score is not an average, the 41 MB-of-native-code lesson, and the id-scheme lesson. 6 tests including synthetic DEX, binary manifest and APK builders. Node 20+, no packages, no network.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.

    Does not decompile code, match malware signatures, or scan native libraries. Manifest parsing is limited to the binary XML string pool without component-level tree parsing.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 18 days ago

    Software
    Node.js 20+

    Permissions

    Read Files
    Listed18 days ago

    What's inside

    Frequently Asked Questions

    Unlike tools that use keyword matching or single-permission triggers, this skill uses a "corroboration" logic that requires both a manifest permission and a corresponding DEX API reference to be present before flagging a risk.

    No, this skill is designed for speed and efficiency; it analyzes the binary manifest string pool and DEX metadata directly without the overhead of a full decompilation or Smali conversion.

    This skill is built for Node 20+ environments and is designed as a dependency-free CLI tool, making it compatible with most AI agents that have terminal or shell execution capabilities.

    You receive the deterministic rule engine, the binary DEX/Manifest parser, a built-in validation test suite, and the logic-gated risk scoring system.

    While it provides a defensible risk score for permissions and API references, it does not perform deep analysis of native libraries (.so files) or signature-based malware detection.

    The output is a structured JSON report that includes a worst-finding risk score, human-readable verdicts, and specific "behavior-keyed" IDs that can be used to link findings across other security tools.

    New in Testing & Debugging