Skillget

Mira

mcp
Guvenlik Denetimi
Gecti
Health Gecti
  • License — License: GPL-3.0
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 107 GitHub stars
Code Gecti
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Gecti
  • Permissions — No dangerous permissions requested
Purpose
This tool is a mobile runtime detection workbench designed for iOS and Android security analysis. It turns mobile app inspection into an AI-native defensive workflow, allowing users to execute live logic and uncover security vulnerabilities within third-party application sandboxes.

Security Assessment
Because it is explicitly designed for mobile penetration testing and runtime manipulation, the tool performs highly sensitive operations. It executes arbitrary shell commands, scripts (JavaScript, Java), and facilitates remote Frida injection to inspect targeted apps. It also makes network requests, operating a local relay server (defaulting to 0.0.0.0) that connects desktop browsers, mobile devices, and AI models over the local network. No hardcoded secrets were found in the scanned code, and no inherently dangerous host permissions were flagged. However, given its extensive dynamic execution capabilities and the requirement to expose a local server endpoint, the overall risk is rated as Medium.

Quality Assessment
The project is actively maintained, with its most recent push occurring today. It is legally transparent and distributed under the GPL-3.0 license. The community trust is currently small but focused, indicated by 34 GitHub stars. Furthermore, a light automated code audit across 12 files detected no dangerous code patterns.

Verdict
Use with caution — strictly within authorized security testing and isolated research environments due to its local network exposure and active runtime manipulation capabilities.
SUMMARY

Mobile runtime detection workbench for AI (iOS and Android)

README.md

English | 简体中文

Mira icon

Mira

Mobile runtime detection workbench for iOS and Android.

AI-native analysis Android and iOS Live logic execution Relay and MCP

Why follow Mira

Mira turns real runtime cases into reusable workflows, tools, and detection notes.

Research Updates

Features

  • App sandbox tools: Use shell, PTY, screen capture, and file tools inside the Mira host app sandbox on Android and iOS.
  • Frida tasks: Run Java and native hooks with RPC for one-time runtime analysis.
  • Relay and MCP: Connect devices through Relay and let AI work in the same sessions through MCP.

Getting Started

  • Relay: PYTHONPATH=. python3 -m mira.relay.server --host 0.0.0.0 --port 8765 --advertise-url http://<your-lan-ip>:8765
  • Browser: Open http://127.0.0.1:8765 on your desktop.
  • Android: Download the latest APK, install it, then enter http://<your-lan-ip>:8765 in the app.
  • iOS: Verified on a real device running iOS 16.7.10. See docs/GETTING-STARTED.md.
  • AI: PYTHONPATH=. python3 -m mira.mcp.server --relay http://127.0.0.1:8765. MCP config: docs/MCP.md.

The APK's offline Python/Frida runtime requires arm64-v8a and supports Android 10 through 15. Android 16 Java hooks are not supported, and no confirmed official Frida fix is available. See the compatibility notes.

Architecture

Contributing

Mira welcomes issues and pull requests from mobile security researchers, reverse engineers, Frida users, MCP users, and device testers.

  • Read CONTRIBUTING.md before opening a focused pull request.
  • Use the issue templates for bugs, security hardening, detection ideas, and device compatibility reports.
  • For security reports, read SECURITY.md first.
  • Scanner-generated hardening PRs are welcome when they include repository-specific reachability reasoning and verification.

Good starting points include native memory-safety review, Android and iOS device testing, Frida workflow examples, MCP client setup notes, and new reusable detection cases.

Live Discovery Examples

Android Remote Frida iOS Remote Frida
Android Remote Frida
Remote shell, runtime inspection, and live Frida execution on Android.
 iOS Remote Frida
Equivalent PTY and Frida workflow adapted to the iOS iSH compatibility layer.
Android LSPosed Trace iOS Jailbreak Trace
Android LSPosed Trace
Construct a Frida path around the app classloader and surface LSPosed traces from runtime state.
 iOS Jailbreak Trace
Ask Claude to roam the live terminal and surface jailbreak-related traces in the device environment.

Public Relay Access

Relay exposed through cpolar

With Relay, you can temporarily expose an authorized session beyond the local network for cloud devices, expert review handoff, and fast evidence sharing.

Research Boundaries

  1. Mira observes and interacts with the Mira host app sandbox.
  2. Mira does not control unrelated third-party apps.
  3. Mira does not provide system-wide remote control.
  4. Mira does not provide root or jailbreak bypass capabilities.
  5. Mira is not a production SDK or a silent background control channel.

Acknowledgements

  • lamda: inspiration for the web workbench interaction model.
  • Termux: Android terminal UX and extensible shell ecosystem.
  • iSH: iOS-side Linux shell compatibility and syscall translation path.

License

GPL-3.0-only.

Yorumlar (0)

Sonuc bulunamadi