Drengr

•

Empowers AI agents to see, interact with, and navigate mobile applications on both Android and iOS devices.

Drengr functions as an MCP server, providing AI agents like Claude, GPT, and Gemini with the essential capabilities to interact with mobile devices. It allows agents to observe screen content, tap, type, and navigate within Android and iOS apps, effectively giving them 'eyes and hands' on mobile platforms. Leveraging annotated screenshots and text mode for efficient communication, Drengr enables AI to understand app states, perform actions, and generate detailed situation reports, enhancing the automation and interaction possibilities for AI agents in mobile environments.

Key Features

010 GitHub stars
02Enables AI agents to see, tap, type, and navigate any mobile application.
03Supports cross-platform mobile interaction on Android and iOS devices, including cloud devices via Appium.
04Offers detailed situation reports after actions, screen exploration with navigation graph mapping, and network capture.
05Provides three core MCP tools: `drengr_look` for screen observation, `drengr_do` for actions, and `drengr_query` for information retrieval.
06Utilizes a vision-first approach with annotated screenshots and an efficient text mode (approx. 300 tokens per screen).

Use Cases

01Developing intelligent mobile copilots that can interact with any installed app.
02Automating complex workflows and tasks within mobile applications using AI agents.
03Facilitating AI-driven testing and quality assurance for mobile applications.