jgm-source/jarvis-opensource

JARVIS: assistente de voz em portugues que ouve, fala e controla o PC. Fork do MARK LV (FatihMakes) com plugins Trello, Cakto, Groq e mais.

What it solves

It gives you a hands-free, JARVIS-style assistant that runs locally and can operate your computer by voice. The original Mark LV project already lets you talk to it, see it as a holographic avatar, and command apps, files, browser, etc. This fork (by João Gabriel) adds Portuguese-language plugins and tweaks aimed at personal and non-commercial use.

How it works

It uses the Gemini Live API for real-time audio streaming, so conversation feels natural. It has a HUD with an animated avatar, and a plugin system: drop a .py file into plugins/ and it learns a new skill. The fork adds many plugins: morning briefing, focus mode, music control, Trello control, parcel tracking, and more. It also adds a cheaper alternative "brain" (using another LLM provider and Edge TTS or Fish Audio) for when you want to save API costs. It keeps your API keys and memory local and never uploads them.

Who it's for

Developers, tinkerers, and voice-assistant fans who want a personal, customizable assistant that runs on Windows (the fork mentions a Windows quick install) and don't mind a non-commercial license (CC BY-NC 4.0). It's especially suited for Portuguese speakers because of the added Portuguese plugins and the developer's welcome message.

Highlights

  • Real-time voice conversation with Gemini Live API, with low latency.
  • Holographic avatar with lip-sync, facial expressions, and status cues.
  • Controls your computer: launch apps, adjust volume, brightness, open URLs, etc.
  • Plugin system: add a single .py file to teach it a new skill.
  • This fork adds 12+ Portuguese plugins (bom dia, trello_control, music_control, etc.) and a cheap "brain" option.
  • Memory and API keys stay local; keys are gitignored and never pushed to GitHub.
  • Model ladder: if one Gemini model fails, it steps to the next measured-ordered rung.
  • Video playback on the HUD: plays YouTube or local videos where the avatar sits, with smart language selection.

Related

  • Project
    ZJU-REAL/Awesome-GUI-Agents

    Awesome-GUI-Agents is a curated list of research papers, datasets, and open-source tools on GUI agents — AI systems that automate computer interfaces. It organizes the field into key modules (perception, exploration, planning, interaction) and tracks recent developments in papers, benchmarks, and reinforcement learning.

  • Project
    Earth-OL-Player/ai_learn_project

    A full-stack AI learning platform that combines a learning roadmap, popular AI interview questions, AI-graded practice with follow-up discussion, a gamified growth system, and a suggestion community — built with Vue 3, Spring Boot, FastAPI, and MySQL.

  • Project
    modelscope/Awesome-Vibe-Research

    A community-curated, openly updatable catalog of AI agents, skills, workflows, and tools for every phase of scientific research — from idea scanning and literature review to experiment execution, paper writing, and post-publication impact analysis.

  • Project
    xpert-ai/xpert

    Xpert is an open-source platform and desktop application for orchestrating AI agent teams, providing a governed environment for building, deploying, and managing complex agentic workflows and business applications.