Report Abuse

Basic Information

NeuralAgent is an open source desktop AI personal assistant designed to perform real work on your computer rather than only converse. It automates mouse and keyboard actions, navigates browsers, fills forms, saves files and runs tasks by combining local Python services and a desktop Electron app with remote-capable language models. The project is structured with a FastAPI backend backed by Postgres, an Electron + React frontend, and a local Python agent daemon that uses pyautogui for foreground automation and a WSL-based option for limited background browser automation on Windows. It supports multiple model providers configured per agent through environment variables and ships modular agent types such as planner, classifier, suggestor, title and summarizer. The README includes setup steps, environment configuration, migration and run commands, and cautions about automated input actions.

Links

App Details

Features
NeuralAgent provides programmatic desktop automation via pyautogui and an AI-driven local agent daemon. It supports background browser automation on Windows through WSL while offering cross-platform support for Windows, macOS and Linux for foreground tasks. The architecture is modular: a FastAPI backend with Postgres for persistence, an Electron desktop wrapper with a React app for the UI, and Python code handling agent actions. Multiple LLM providers are supported per agent including OpenAI, Azure OpenAI, Anthropic, AWS Bedrock, Gemini and Ollama. Agent types are configurable per model and include PLANNER_AGENT, CLASSIFIER_AGENT, SUGGESTOR_AGENT, TITLE_AGENT, COMPUTER_USE_AGENT and SUMMARIZER_AGENT. The repo contains environment-driven configuration, alembic migrations for DB setup, example env files, demo walkthroughs and a note about optional internal screenshot logging disabled for open-source use.
Use Cases
NeuralAgent helps users automate repetitive desktop and web workflows by letting language models plan and execute clicks, typing and browser navigation, which can save time on data entry, file creation, and multi-step tasks. Its modular agents separate concerns so planning, classification, summarization and UI interactions can be customized or swapped to match workflows. Support for multiple LLM providers and per-agent model configuration enables using cloud or local models according to cost, latency or privacy needs. The app runs locally with a backend database for state and can be extended by developers through its Python and JavaScript components. The README provides clear installation, environment and startup instructions and warns users to test responsibly because the tool moves the mouse and types on behalf of the user.

Please fill the required fields*