yashab-cyber

yashab-cyber / opendroid

Public

Your Open Autonomous Android Agent — A production-ready, self-planning AI assistant powered by local/remote LLMs and accessibility-driven screen automation.

accessibility ai-agent android automation autonomous
10
5
94% credibility
Found May 24, 2026 at 15 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Kotlin
AI Summary

OpenDroid is an Android app that acts like a smart, voice-controlled personal assistant living on your phone. Instead of just answering questions, it can actually do things for you — sending messages, making calls, setting alarms, controlling your phone settings, ordering food, booking rides, and much more. You talk to it naturally with your voice, and it breaks down your requests into steps, executes them one by one, and adapts if something goes wrong. It also remembers things about you over time, so it gets better at helping with your specific needs. The app works with various AI services to understand what you want and how to accomplish it.

How It Works

1
📱 You download OpenDroid from GitHub

You find this app online and install it on your Android phone, just like any other app.

2
⚙️ You connect your AI brain

In the settings, you enter your AI service account so the assistant can think and plan tasks for you.

3
🎙️ You say 'Hey OpenDroid' and it wakes up

The assistant starts listening for your voice command, ready to help you with whatever you need.

4
🗣️ You tell it what to do

You say something like 'Check if it's going to rain, and if so, text my wife that I'll be late and set an alarm for 6 PM.'

5
🧠 Your assistant figures out the steps

OpenDroid breaks your big request into smaller tasks and shows you the plan before doing anything.

6
OpenDroid works its magic
✅
Everything goes smoothly

The assistant completes all the steps and tells you it's done.

🔄
Something needs adjusting

If a step doesn't work, the assistant automatically figures out a different way to get it done.

✨ Your task is complete

Your wife got the message, your alarm is set, and OpenDroid remembers this for next time.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 15 to 10 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is opendroid?

OpenDroid is an autonomous AI agent for Android that plans and executes tasks on your device using natural voice commands. Instead of acting as a chatbot, it breaks complex goals into sequential steps, executes them through direct device controls and accessibility automation, and dynamically replans when something fails. Built in Kotlin with Jetpack Compose, it connects to multiple LLM providers (OpenAI, Claude, Gemini, Groq, Mistral, Ollama, and others) or runs entirely offline via local models. It includes a multi-tier memory system that retains context across sessions, a vision engine that reads screens, and a voice interface with wake word detection.

Why is it gaining traction?

The hook is clear: this is a genuinely autonomous agent, not a wrapper around an API. It handles compound multi-step commands like "check if it will rain, and if so, text my wife that I'll be late" by decomposing, executing, and adapting in real-time. The action library is surprisingly broad -- system controls, calls, SMS, WhatsApp automation, calendar, media playback, smart home, UPI payments, food delivery, and ride booking all work through intent fallbacks when direct APIs aren't available. The LLM hallucination defense layer that auto-corrects unknown action names is a thoughtful touch that improves reliability. Local Ollama support means privacy-sensitive users never have to send data to external APIs.

Who should use this?

Android power users who want hands-free voice control over their device. Developers exploring autonomous agents on mobile will find a well-structured reference implementation. Privacy-conscious users who want AI assistance without cloud dependencies. Anyone tired of manually navigating apps for repetitive tasks. Not ready for enterprise deployment, but compelling as a personal assistant or development sandbox.

Verdict

The architecture is solid (clean layers, Hilt DI, Room persistence, Compose UI) and the feature set is ambitious. However, with only 10 stars and a 0.95% credibility score, this is early-stage software that needs community validation before production use. The README is thorough and the build setup is well-documented, but test coverage and stability are unknown quantities. Worth watching -- clone it, run it on a test device, and see if it handles your use cases before committing to it.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.