MuseLite

agent
Guvenlik Denetimi
Uyari
Health Uyari
  • License — License: GPL-3.0
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 7 GitHub stars
Code Gecti
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Gecti
  • Permissions — No dangerous permissions requested

Bu listing icin henuz AI raporu yok.

SUMMARY

MuseLite is an open-source AI workspace combining chat, Agent tools, browser automation, and scheduled tasks for users who enjoy Muse and Dots.

README.md

MuseLite

MuseLite is a Python-first AI workspace for Android. It combines an OpenAI-compatible chat client, persistent conversations, an execution sandbox, browser tools, scheduled prompts, and optional phone automation in one native Android application.

Features

  • Chat with any OpenAI-compatible HTTPS provider.
  • Persistent conversations and model settings stored in SQLite.
  • API keys encrypted with the Android Keystore.
  • Scheduled prompts: type a prompt, choose a time, and let MuseLite send it in the selected conversation.
  • Agent tool loops for multi-step work, with progress and stop controls.
  • A persistent Alpine Linux sandbox powered by PRoot for shell commands and workspace files.
  • Browser tools for fetching pages, managing tabs, and saving useful results into the workspace.
  • Optional Android phone control through the accessibility service and a foreground task service.
  • Safe Markdown rendering for assistant messages.
  • Android 11+ arm64 support, with a desktop Toga preview for development.

Requirements

  • Windows x86-64 for the provided build script.
  • Python 3.13.
  • Internet access on the first build to download Briefcase, the Android SDK, Gradle, and Maven dependencies.
  • An Android 11+ arm64 device or emulator for the packaged application.

WSL, Linux, and administrator privileges are not required by build_windows.ps1.

Build the APK on Windows

From the repository root, run:

powershell -ExecutionPolicy Bypass -File .\build_windows.ps1

The script creates a local .venv, downloads build tools into cache, stages the Android project, and writes a debug APK to dist. On the first run, accept the Android SDK licenses when prompted.

Install the newest APK on a connected device with:

$apk = Get-ChildItem .\dist\*.apk |
    Sort-Object LastWriteTime -Descending |
    Select-Object -First 1
& .\cache\tools\android_sdk\platform-tools\adb.exe install -r $apk.FullName

Debug APKs are intended for development and testing. They are not release-signed packages for an app store.

Configure a model

Open Settings after launching MuseLite and enter:

  1. An OpenAI-compatible HTTPS base URL.
  2. A model ID.
  3. An API key.

For example, DeepSeek can use https://api.deepseek.com with the model ID deepseek-flash. The provider endpoint is validated before settings are saved. The API key is kept in the app's private storage and encrypted with Android Keystore.

Schedule a prompt

  1. Tap Start New Conversation.
  2. Type the prompt you want MuseLite to send later.
  3. Tap Schedule beside the input field.
  4. Choose the hour and minute, then tap Confirm.

MuseLite returns to the main screen after confirmation. At the selected time, Android opens the saved conversation and sends the prompt that was in the input field when the task was created. An empty draft is not saved as a recent conversation.

Use Agent phone control

Phone control is disabled by default. To enable it:

  1. Open Settings > Phone Control.
  2. Enable the MuseLite accessibility service in Android system settings.
  3. Grant notification permission when Android requests it.
  4. Turn on Allow Agent to operate the phone in MuseLite.

When enabled, the Agent can inspect the current screen, tap or long-press, type or clear text, swipe, press Back/Home/Recents, open an app by package name, and wait for a target to appear or disappear. Tasks run in a foreground service with a persistent notification and can be stopped from the notification or the chat. The Agent does not regain access after the process is killed.

Local sandbox

The first shell_execute call unpacks the bundled Alpine root filesystem. The persistent paths are:

  • /var/muselite/workspace for user files.
  • /var/muselite/memory for saved notes and memory.

The minimal image does not include Python or pip. If needed and network access is available, install them inside the sandbox with:

apk add --no-cache python3 py3-pip

The installed packages and files remain in the app's private data directory.

Project layout

src/muselite_py/          Python UI, Agent, storage, tools, and sandbox
java/com/muselite/python/ Android bridges, services, scheduling, and phone control
template/                 Briefcase Android template and native resources
scripts/                  Windows build, staging, verification, and asset tools
vendor/                   Reproducible Alpine and PRoot inputs
tests/                    Python unit tests

Development checks

Run the Python test suite from the repository root:

$env:PYTEST_DISABLE_PLUGIN_AUTOLOAD = '1'
python -m pytest -q tests

For Android validation, install the generated APK on an Android 11+ arm64 device and verify launch, streaming chat, scheduled prompts, multi-step tool calls, sandbox commands, persistence after restart, browser tools, and phone-control authorization changes.

Native components and licenses

The bundled PRoot loader, PRoot binary, and Alpine root filesystem under vendor/ are derived from the OpenMinis Android 1.13 release. See LICENSE and THIRD_PARTY_LICENSES.md for project and third-party license information.

MuseLite is released under the GNU General Public License v3.0. See LICENSE for the full text.

Yorumlar (0)

Sonuc bulunamadi