aaswordman/operit

★ 4,596KotlinAudience · vibe coderComplexity · 4/5Setup · moderate

Mindmap

mindmap
  root((Operit))
    What it does
      AI controls Android phone
      40+ built-in tools
      On-device Linux environment
    AI backends
      OpenAI and Claude APIs
      Local llama.cpp models
      MNN on-device inference
    Tool categories
      File management
      Web and network
      Screen automation
      Terminal commands
    Features
      Memory across sessions
      Custom AI personas
      MCP plugin marketplace
      Voice interaction

mindmap root((Operit)) What it does AI controls Android phone 40+ built-in tools On-device Linux environment AI backends OpenAI and Claude APIs Local llama.cpp models MNN on-device inference Tool categories File management Web and network Screen automation Terminal commands Features Memory across sessions Custom AI personas MCP plugin marketplace Voice interaction

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Let an AI model manage files, install apps, and automate on-screen interactions on your Android phone using accessibility services or ADB

USE CASE 2

Run Python or Node.js scripts on your Android phone through the built-in Ubuntu 24 Linux environment without a PC

USE CASE 3

Use a locally downloaded AI model offline on your Android device for full privacy using the MNN or llama.cpp backend

USE CASE 4

Install additional AI tool packages from the built-in MCP marketplace to expand what the AI can do on your device

Tech stack

KotlinAndroidMNNllama.cppPythonNode.js

Getting it running

Difficulty · moderate Time to first run · 30min

Requires Android 8.0 or later with at least 4 GB of RAM, accessibility or ADB permissions needed for automation features.

No license information is mentioned in the explanation.

In plain English

Operit is an AI assistant app for Android that goes well beyond simple chat. Rather than just answering questions, it connects an AI model to your phone's actual capabilities, letting it take actions on your behalf such as managing files, making web requests, installing apps, running terminal commands, and interacting with other apps on screen by tapping and typing automatically. The app ships with more than 40 built-in tools organized into categories. There are file system tools for reading, writing, searching, and converting files. There are network tools for fetching web pages and downloading content. System tools can manage app permissions and automate on-screen interactions through three different modes: Android's accessibility services, ADB debugging, or root access depending on what is available on the device. There is also a built-in Ubuntu 24 Linux environment, meaning you can run Python scripts, Node.js, package managers, and complex command-line tasks directly from the app on your phone. Operit supports running AI through external API providers like OpenAI, Claude, Gemini, and OpenRouter. It also supports running AI models locally on the device using the MNN or llama.cpp frameworks, which means the AI can work entirely offline once a model is downloaded, keeping your data on the device. The app has a memory system where the AI automatically categorizes and stores information from conversations so it can remember your preferences and past context across sessions. You can define custom AI personas using character cards, and multiple personas can have separate conversation histories and even chat with each other. A plugin marketplace called the MCP/Skill market lets you extend what the AI can do by installing additional tool packages. Voice interaction is supported with both local and cloud-based text-to-speech and speech-to-text options, and voice or specific audio can be used to wake the app. The app requires Android 8.0 or later and at least 4 GB of memory. Installation is done by downloading the APK from the official releases page. The README is primarily in Chinese, with an English version available in a separate file.

Copy-paste prompts

Prompt 1

How do I set up Operit on Android to use a local AI model with llama.cpp so it works without an internet connection?

Prompt 2

Show me how to automate on-screen taps and app interactions on Android using Operit's accessibility service mode

Prompt 3

How do I run a Python script on my Android phone using Operit's built-in Linux environment?

Prompt 4

How do I connect Operit to the Claude or OpenAI API as the AI backend instead of a local model?

Prompt 5

What built-in tools does Operit give the AI by default and how do I add more from the MCP marketplace?

Open on GitHub → Explain another repo

← aaswordman on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.