Skip to content
KitploitKITPLOIT
ToolsBlog
Submit
ToolsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

··Feeds·Contact·Privacy·© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
HydraDragonAV-Mobile — Android Antivirus which doesn't require root, adb, ca install and cloud with many features and ways to detect more zero-day malware | Kitploit
Tools/GitHubGitHub/hydradragonantivirus/hydradragonav-mobile
Android SecurityDefensive ToolsStatic AnalysisDynamic Analysis (Sandboxing)PhishingNetwork SecurityMalware AnalysisMobile SecurityMachine Learning

Most Popular

View all →

Discover the most used tools by our community.

Explore all tools

Browse our collection of tools

View all tools →
Share
GitHubhydradragonantivirus/hydradragonav-mobile

HydraDragonAV-Mobile

Android Antivirus which doesn't require root, adb, ca install and cloud with many features and ways to detect more zero-day malware

View RepositoryWebsite
161510 days agoNot yet reviewed

🐉 HydraDragonAV Mobile

Advanced Android Antivirus With Threat Protection

⚠️ Project on hold. This project did not get enough interest, so I have paused it for a while. Development has moved to HydraDragonAntivirus.

Discord Community Server: https://discord.gg/7XMCuj5mbP

HydraDragonAndroid (Logo made by discord user called: 1tophbeifong)

Android Java Rust Security Languages License

HydraDragonAV Mobile is a multi-layered Android Antivirus and Security suite combining static analysis (YARA-X + ClamAV signatures + code anomaly detection), dynamic behavior analysis, and a lightweight on-device ML classifier — all gated by a NSRL-backed whitelist so known-good software is never a false positive. Designed with a Zero-Trust architecture, it actively defends the device against ransomware, clickjacking, spyware, SMS scams/phishing, and zero-day threats. When a flagged malware app is launched, it is automatically force-stopped and blocked in real time.

📋 Requirements: Android 8.0 Oreo (API 26) or newer. The per-app dynamic network analysis — attributing each DNS/connection to the exact app that made it and feeding it to the YARA-X hydradragon module — relies on ConnectivityManager.getConnectionOwnerUid, which is only available on Android 10+. On Oreo/API 26-28 the rest of the suite still runs, but per-app connection attribution is unavailable.

🤖 Android-only detection scope. This is a minimalist, Android-focused fork of the original HydraDragonAntivirus — it only detects Android malware. The native scan engine only runs its ClamAV/YARA-X signature matching on files it can confidently identify as an Android-relevant type (APK/ZIP, DEX, ELF, HTML, ASCII text, PDF, images); desktop-only formats (Windows PE, OLE2/Office, Mail, Mach-O, Java) and files of an indeterminate type are skipped entirely, rather than scanned with a signature database that was never meant for them. For more details, see the wiki (ClamAV Integration, Known Limitations).

🚀 Key Features

  • ⚡ Photon Technology: Ultra-fast, multi-threaded scanning engine utilizing ConcurrentHashMap caching to instantly re-verify previously scanned safe applications with zero CPU overhead.
  • 🧠 Native Rust Scan Engine: A JNI-bridged Rust core (libhydradragonandroid.so) does the heavy lifting — YARA-X + ClamAV signature matching, archive extraction (zip/gz/tar/xz/lzma/7z/rar, including nested APKs), AXML manifest parsing, and dangerous-permission counting, all straight from bytes in memory (no temp files required).
  • 🧠 Burn ML Binary Classifier: A tract-based Burn malware/benign classifier scores every scanned APK with a confidence value (0.0–1.0). Trained on both malware and benign datasets for accurate binary classification. A lightweight logistic-regression classifier (AIEngine) also scores DEX-level behavior (obfuscation, dynamic loading, crypto/socket/shell APIs, adware SDKs).
  • 🛡️ Ransomware & Screen-Locker Mitigation: Real-time on-screen text detection (via Accessibility Service + OCR) recognizes ransom notes and forcefully terminates screen-locking ransomware, in 20 languages.
  • 📵 SMS Scam & Phishing Detection: The same multi-language on-screen text scanner catches smishing lures (fake account verification, prize scams, parcel-delivery scams, OTP requests, bank alerts) wherever they're rendered — Messages app, notification previews, or a spoofed WebView — without ever reading your SMS inbox directly.
  • 🖥️ Live Screen OCR: A MediaProjection-based capture service periodically OCRs the foreground screen and feeds the extracted text into the native threat scanner, catching scams that only ever appear as rendered pixels.
  • 🖼️ MalwareFoundActivity as Dialog: When a threat is detected during a background scan, a floating dialog overlay (not a full-screen activity) alerts the user without disrupting their current task.
  • 🚫 UI Hijacking / Clickjacking Protection: Detects and blocks automated rapid-click permission-granting attacks, notification-spam floods, and repeated overlay/dialog abuse from malicious apps. Task-hijacking (StrandHogg activities-count check) and a screen-recording/FLAG_SECURE guard are also built in as opt-in Settings toggles (the FLAG_SECURE guard can be intentionally disabled via an "Allow screen recording" toggle — tamper detection still catches unauthorized removal).

🛠️ Technical Architecture

HydraDragonAV Mobile operates across four core pillars:

  1. GuardService: A persistent Foreground Service acting as the brain of the active defense system — watches the Downloads folder in real time and orchestrates scans, monitoring threats 24/7.
  2. DynamicAnalysisService: An Accessibility Service that prevents automated UI hijacking (Clickjacking), blocks overlay/notification-spam attacks, and scans on-screen text for ransomware/SMS-scam/phishing wording in 20 languages.
  3. ScreenCaptureService: A MediaProjection-based OCR service that periodically captures and reads the foreground screen, feeding extracted text into the native scanner for live threat detection.
  4. ScanEngine + NativeScanner: The Java orchestration layer and its Rust-native counterpart (libhydradragonandroid.so) — evaluating X.509 certificates, SHA-256 hashes, dangerous permissions, app install sources, YARA-X/ClamAV signatures, and ML anomaly scores.

A local DnsVpnService additionally filters DNS lookups against known-malicious domains, and a NetworkSecurityScanner/NetworkMonitor pair watches live connections for MITM interception, ARP spoofing, and malicious IP/C2 traffic.

🚫 Why Not Shizuku (or Similar Shell-Privilege Frameworks)?

Tools like Shizuku grant an app ADB-shell-level privileges (process killing, PackageManager operations beyond the public API, real iptables-style firewalling, etc.) without full root. It's a legitimate, useful project — just not the right fit inside a production antivirus:

  • It doesn't remove the attack surface, it relocates it. Shizuku's own privileged service is reachable via Binder by any app that requests it; embedding a shell-privilege-escalation client inside an AV means a compromise of the AV process (or a bug in how it talks to Shizuku) inherits ADB-level reach across the device — the exact blast radius an AV is supposed to shrink, not grow.
  • It can't self-activate in production. Shizuku needs the user to start it every boot via adb shell (wireless or USB debugging) or root, each time, unless the device is already rooted — completely impractical for a mass-market install, and this app already refuses to run at all on rooted devices (RootCheck) and warns the user when USB/wireless debugging is left on (DebugModeCheck/DebugModeWarning) precisely because both states widen the attack surface.
  • The marginal capability isn't worth the trust model change. The main draw (a real packet-level firewall, force-stopping arbitrary apps) is already reasonably covered here by VpnService-based DNS/domain/IP filtering (Web Shield) without ever requiring the user to authorize shell access to a third-party broker service.

In short: a security product that asks the user to first grant it debug-shell reach is solving its threat model backwards. The app's actual privilege escalation surface is intentionally kept to what Android's public APIs (AccessibilityService, VpnService, PackageManager) allow — nothing that requires ADB, root, or a privileged companion app.

🧬 Why Was TLSH Removed?

Early builds shipped a TLSH fuzzy-hashing engine that compared every scanned APK/ELF/DEX against a MalwareBazaar-derived digest database. It was removed because it was the slowest, heaviest, and least useful engine in the whole suite:

  • It was the slowest engine by far. Telemetry showed TLSH consuming ~49 seconds of single-threaded wall time per scan — roughly 92% of the entire scan scope — while ClamAV (4 workers, accumulated) and YARA-X finished their passes in a fraction of that. The bottleneck wasn't hashing, it was a linear diff search over a ~177K-digest ELF database for every scanned buffer.
  • It repeated Xvirus's mistake — a fake "100% detection" in tests but not real World. Pulling a database from well-known malware-collection sites and matching everything against it produces impressive-looking but meaningless numbers: the detector can only ever re-flag exactly what was already downloaded, giving a false sense of total coverage. In practice TLSH added only ~0.1–0.5% redundant detections on real-world samples — everything it flagged was already caught by the ClamAV/YARA-X/ML layers.
  • It was bloated. The bundled TLSH reference databases (ELF, APK, DEX) shipped megabytes of digests into the APK for a layer that contributed next to nothing beyond what signature and ML engines already deliver.

The features that actually earn their place — YARA-X + ClamAV signatures, the ML classifier, NSRL whitelisting, emulation, and URL/IP filtering — cover the same ground without the runtime cost or the empty promise of "100%".

Designed with 💻 by @elnureisayeva1-cloud (creator) & @Siradankullanici (backend development)

Note: the maximum file size the engine will scan is user-configurable from Settings (10–2048 MB, default 650 MB) — files above the limit are skipped entirely rather than partially scanned.

⚙️ How to Build & Install

  1. Clone the repository (submodules included — the native engine vendors a custom YARA-X fork with the hydradragon scanning module):

    root@kitploit:~
    git clone --recurse-submodules https://github.com/HydraDragonAntivirus/HydraDragonAV-Mobile.git
    cd HydraDragonAV-Mobile
    
  2. Build the native Rust engine for Android (see hydradragonandroid/README.md for prerequisites):

    root@kitploit:~
    cd hydradragonandroid
    build-android.cmd
    
  3. Build the Android app as usual (Gradle) — the native .so output is picked up automatically from app/src/main/jniLibs/.

🗄️ Data Pipeline (Whitelists, Signatures & Filters)

The on-device detection assets are generated offline from public threat-intel and NSRL sources, then bundled into the app:

  • gen_whitelist_packages.py — builds whitelist_packages.db (SQLite), a full-detail NSRL Android package whitelist (name, version, manufacturer, OS, hashes) joined from the NSRL RDS database.
  • gen_whitelist_apk.py — extracts whole-APK NSRL MD5 hashes for the native binary-fuse XOR whitelist filter.
  • gen_ip_lists.py / build_url_xfilters.py / build_xfilters.sh — build the malicious IP/URL/domain filters (XOR-filter based) used by the native IP/URL threat scanners.
  • clam_juice.py — filters ClamAV signatures to keep only Android-relevant platforms (Andr, Unix, Linux, Email, PUA) plus all Phishing signatures; excludes Win/Osx/Java:
    root@kitploit:~
    python clam_juice.py --directory database_non_filtered --output database_filtered --profile cross-platform
    

📦 Dataset Pipeline

  • fdroid-downloader.py — downloads benign APKs from the official F-Droid repository for the dataset pipeline. Only downloads APKs not already present locally; compares by filename so updated versions are picked up automatically.
  • dataset/benign/F-Droid/ — collected F-Droid open-source APKs (benign by nature with 2805 APKs). https://huggingface.co/datasets/HydraDragonAntivirus/BenignAPKDataset/tree/main (uploaded via "hf upload-large-folder HydraDragonAntivirus/BenignAPKDataset dataset\benign --repo-type=dataset --num-workers=4")
  • dataset/malware/ — malware samples (stored separately, not in this repo. Link: https://drive.google.com/file/d/1Il2bvcjxmc0w-XirL27R0_bKpznIeIiG Password: infected).
  • yarGen (YARA rule generation) dataset: https://drive.google.com/file/d/1BrLoMNzz2pJB886A6JfEqFris_c7Ua8o
  • Note: yarGen contains data2 = PE benign dataset javascript = javascript benign dataset android = andorid benign dataset

Donate:

  • For donate HydraDragonAV-Mobile creator go to here:https://boosty.to/yusif_musayev/single-payment/donation/817954/target?share=target_link (Yusif Musayev HydraDragonAV-Mobile creator)
Download Tool
  • 🔍 On-Install & On-Demand Scanner: Quick (installed apps only) or Full (entire storage including SD card) scans; a BroadcastReceiver intercepts and scans newly installed/updated APKs automatically. Background scans can be adopted by the UI if the user opens the scan screen mid-scan — no redundant re-scan.
  • 🎯 Active Threats Counter: A dedicated "ACTIVE THREATS" stat on the scan screen shows the number of currently-detected threats alongside the scanned/threat totals.
  • 📋 Rich Threat Dialogs: Tapping a threat shows all detection reasons in a bulleted list with clickable VirusTotal URLs for immediate verification, plus Destroy / Ignore / Ignore-Signature actions.
  • 🌐 Network Security Monitor: Tracks live connections, flags known-malicious/C2 IPs (validated as public IPv4 before scanning — private/loopback/multicast addresses are rejected) and anonymizer/tunnel-service-shaped domains (dyndns, ngrok, .onion lookup attempts), detects MITM/TLS interception via untrusted CAs, and can spot ARP spoofing on the local network. (Domain/DNS-based — this observes plain-DNS lookup patterns, not actual Tor circuit traffic, which never touches the device's DNS resolver.)
  • 🌐 DNS-Filtering VPN (Web Shield): A local, on-device VPN service blocks resolution of known-malicious/phishing domains — no traffic proxying, decryption, or inspection, only DNS lookups are filtered.
  • 🗃️ NSRL-Backed Whitelisting: Known-good software is cleared through two independent NSRL-derived layers — a binary-fuse XOR filter of whole-file SHA-256 hashes (native, in-memory) and a SQLite database of full NSRL package metadata (name, version, manufacturer, OS) — so a legitimate app is never flagged, while a malicious hash can never "borrow" a whitelist entry by luck.
  • 🔒 Zero-Trust Mode: Optional stricter mode where any app that merely survives every detector (rather than being explicitly cleared) is still flagged as suspicious, with a full audit trail.
  • 🧠 Behavioral Detection Suite: A dedicated set of runtime detectors, each individually toggleable from Settings — UI/notification-spam (adware), a device-became-rooted-mid-session monitor, a combined permission+suspicious-DNS-pattern risk score, and ransomware behaviour detection (see below). Every hit immediately kills the offending app's background process (where possible) and pops the system uninstall prompt right away, instead of waiting for the next scheduled scan.
  • 🪤 Ransomware File Traps: When a freshly-installed, still-unvetted app is granted "All Files Access", HydraDragon temporarily drops a decoy file (named to sort first in a typical alphabetical encryption pass) into Downloads/Documents/Pictures/DCIM. Any rename, content change, or deletion of that file — something no legitimate app has a reason to ever touch — is 100% certain ransomware behaviour, not a heuristic guess. Traps auto-expire within 24 hours so they never clutter a user's files.
  • 🔁 Rename-Burst Ransomware Detection: Independently, a burst of files being renamed with an appended suffix (whatever that suffix actually is — no hardcoded ".enc"/".locked" list) shortly after an app gains file access is flagged as in-place encryption, the filesystem-level shape every ransomware family shares regardless of its specific extension.
  • 🐴 Native-Code Emulation (Unicorn Engine): Runs an embedded native library's JNI_OnLoad/entry code in a bounded, syscall-free CPU sandbox (ARM/ARM64/x86/x86_64) to reveal strings — like a C2 URL — that a decode/decryption routine only produces at runtime, never as static plaintext. Fully toggleable from Settings.
  • 🎣 Malicious/Phishing URL String Scanning: Extracts every embedded http(s):// URL from a scanned file's raw bytes (APK or otherwise) and checks it against the native malware/phishing URL xor filters — a full URL (with path) is far more specific than a bare domain, meaning fewer false positives and more precise detections than domain-only matching.
  • 🔐 Settings Self-Protection: Every Settings toggle/button is hardened against non-human tampering — an overlay-based tapjacking attempt is blocked via Android's own FLAG_WINDOW_IS_OBSCURED check, and an inhumanly fast burst of setting changes (the signature of a malicious accessibility service driving the UI directly) is detected, reverted, and blocked, so only the actual device owner can ever change protection settings.
  • 🧹 Bloatware Cleaner, Self-Protection & Root Detection: Rounds out the suite with background-app cleanup, Device Admin tamper/uninstall resistance, and rooted-device detection.
  • 🌍 Multilingual Support: Fully localized UI and threat-detection keyword lists across 20 languages: English, Turkish, Spanish, German, French, Russian, Portuguese, Arabic, Italian, Dutch, Polish, Ukrainian, Chinese (Simplified), Japanese, Korean, Hindi, Indonesian, Vietnamese, Persian, and Thai.