Devotd Github
NewAI agents

wangzan101/mika v0.1.2 — On-device AI agent for Android

wangzan101/mika is an on-device AI agent that manipulates apps and the web on Android devices and uses a local Linux sandbox; the latest release is v0.1.2.

Repositorywangzan101/mika ↗

wangzan101/mika is an on-device AI agent that manipulates apps and the web on Android devices and uses a local Linux sandbox; the latest release is v0.1.2.

GitHub description: Mika — 你手机上的私人 AI 助理,替你操作 App、上网、把事办成。An on-device AI agent for Android that operates apps and the web to get things done for you.

stars
★ 125
Forks
22
Language
Kotlin
License
Apache-2.0
Latest release
v0.1.2
Read original source ↗View on GitHub ↗

wangzan101/mika is an on-device AI agent that operates apps and browses the web on Android phones. The latest release published in the repository is v0.1.2. The provided information does not list anything newly changed in this version, so specific changes cannot be confirmed. The repository description and README present app manipulation, web browsing, and running a local Linux sandbox as the main features. The project is written in Kotlin and uses the Apache-2.0 license.

Work is carried out in a separate background screen so that users can keep using the phone screen. App manipulation uses Android accessibility features, and steps that require user intervention, such as login, authentication, and payment, are handed back to the user. The Linux sandbox is an isolated environment that runs inside the device. According to the README, users configure their own model provider keys, requests are sent to that provider, and there is no Mika server. Per-app usage knowledge and encrypted personal memory are also described as being managed on the device.

The research record rates the technical depth and implementation quality highly, citing multiple platform capabilities and a large test tree. However, this evaluation is a record based on the repository, and the provided information contains no performance measurements or actual run results. No change history for v0.1.2 is presented either. Therefore, what can be confirmed at present is the scope of the feature design, and the grounds for judging the performance and success rate of real tasks are limited.