| Demo | Quick Start | Release | Issues | CI Builds | Debug Release | English | 简体中文 |
Unlike traditional mobile AI chat apps, OpenOmniBot runs directly on your device and can operate your Android phone like a human, including apps, gestures, and system settings.
OpenOmniBot is an on-device AI agent built with native Android Kotlin and Flutter. Instead of stopping at chat, it focuses on the full loop of understand -> decide -> execute -> reflect.
- Extensible tool ecosystem: Skills, Alpine environment, browser access, MCP, and Android system-level tools.
- Phone task automation: Uses vision models to understand and operate mobile interfaces.
- System-level actions: Supports scheduled tasks, alarms, calendar creation/query/update, and audio playback control.
- Memory system: Short-term and long-term memory with embedding support.
- Productivity tools: Read and write files, browse the workspace, use the browser, and access the terminal.
Open the settings page from the left sidebar:
Then open the scenario model settings:
Note: Memory embedding requires an embedding model. For the best overall experience, the other scenarios should use multimodal or vision-capable models whenever possible.
The app usually initializes the Alpine environment automatically on startup, and you can also manage that environment from the same settings area.
You can ask OmniBot to install a skill by simply sending it the repository link. Recommended collection: https://github.com/OpenMinis/MinisSkills
Enable or disable skills from the skill repository:
Before starting a task, open the chat page and grant all required permissions from the top-right corner.
Supports both MNN and llama backends.
Scheduled tasks can execute work such as VLM tasks and subagent flows. Alarms are reminder-only. A subagent can be assigned a complete task and behaves like a full agent.
- Flutter SDK
3.9.2+ - JDK
11+
git clone https://github.com/omnimind-ai/OpenOmniBot.git
cd OpenOmniBot
# Required only when building the full omniinfer edition.
git submodule update --init third_party/omniinfer
git -C third_party/omniinfer submodule update --init framework/mnn
git -C third_party/omniinfer submodule update --init framework/llama.cpp
cd ui
flutter pub getIf Flutter reports Could not read script '.../ui/.android/include_flutter.groovy', run:
flutter clean
flutter pub getcd ..
# Slim standard edition, without local inference
./gradlew :app:installDevelopStandardDebug -Ptarget=lib/main_standard.dart
# Full omniinfer edition, with local inference
./gradlew :app:installDevelopOmniinferDebug -Ptarget=lib/main_omniinfer.dart
OpenOmniBot/
├── app/ # Android host app: entry point, agent orchestration, system abilities, MCP, services
├── ui/ # Flutter UI: chat, settings, tasks, memory, and web chat bundle
├── baselib/ # Shared core libraries: database, storage, networking, model config, OCR, permissions
├── assists/ # Automation engine: task scheduling, state machine, visual detection, execution control
├── accessibility/ # Accessibility and screen perception: accessibility service, screenshots, projection
├── omniintelligence/ # AI abstractions: model protocol, task status, request/response models
├── uikit/ # Native overlay UI: floating ball, overlay panels, half-screen surfaces
├── third_party/omniinfer/ # Local inference runtime and Android integration modules
└── ReTerminal/core/ # Embedded terminal experience modules
APK 由 GitHub Actions 自动构建，每次 push 到 main 分支触发。 Artifact 保留 7 天，过期后需重新 push 触发构建。 点击链接跳转到 Actions 页面下载 zip，解压后得到 APK。
| 构建时间 | Commit | 提交信息 | OmniInfer 版 | Standard 版 |
|---|
使用 debug keystore 签名的 Release 版 APK，适合测试安装。
⚠️ 每次构建 keystore 不同，无法覆盖安装（需先卸载旧版）。
- 手动触发：进入 Actions → Debug Release APK，点击 "Run workflow"
- Tag 触发：推送
debug-v*tag（如git push origin debug-v0.5.2.6）
构建产物发布在 Releases 页面。
Thanks to the community （including LINUX）developers supporting OpenOmniBot.
Special thanks to these open-source projects:
 |