Mira
A mobile runtime detection workbench for iOS and Android that runs Frida-based hooks and shell tools inside a host app sandbox, and lets an AI work in the same sessions through MCP.
Mira is a workbench for mobile security researchers and reverse engineers studying runtime behavior on iOS and Android. Inside its host app sandbox it gives access to a shell, a pseudo-terminal, screen capture and file tools on the device, and it can run Java and native Frida hooks with remote procedure calls for one-off runtime analysis.
A relay server connects a browser or the Mira Android or iOS app to a device over the network, and a Model Context Protocol server lets an AI agent join the same sessions to help drive the analysis. The project publishes ongoing research updates, such as detecting root, emulators and screen-projection tools through side channels in process-audit logs, turning each real case into a reusable workflow, tool or detection note.
The Android app's offline Python and Frida runtime needs an arm64-v8a device and supports Android 10 through 15; Android 16 Java hooks are not supported. It is licensed under the GPL-3.0 and welcomes contributions from mobile security researchers, Frida users and device testers.