Repolode
Submit
deepnovacore

deepnovacore/NovaPersonalAgent

Open-source personal AI agent and voice assistant with a dual-brain design, personal memory, and delegation to coding agents.

View on GitHubWebsite
Stars
124
Forks
38
Open issues
6
Last commit
yesterday

What is it?

NovaPersonalAgent is a desktop and headless-server personal agent with a voice interface. A front brain keeps the conversation responsive, while a back brain runs longer tasks in the background. Users can ask Nova to clarify goals, request approval for permissions, and check results against their own criteria before taking over a task.

Nova builds personal memory from sources the user connects, such as local folders, mail, calendars, conversations and documents. Memories can be traced to their sources, edited or removed. It also offers Todos, Ideas, Goals, Feeds and a Profile that turn that context into suggestions the user can adopt or ignore.

The project supports configurable ASR, LLM, TTS and vision providers, MCP tools, and coding backends including Codex, OpenCode, CodeBuddy, Pi and DeepSeek Harness over ACP. It runs on macOS arm64, Windows x64 and Ubuntu 22.04+ x64, and an iPhone client can pair with a desktop or server over Tailscale. Installation is via npm, and the project is written in TypeScript.

Key features

  • 01Dual-brain design: responsive conversation with background task execution.
  • 02Personal memory built from user-connected folders, mail, calendars and documents, with traceable sources.
  • 03Todos, Ideas, Goals, Feeds and Profile that offer suggestions the user can adopt.
  • 04Voice interface with configurable ASR, LLM, TTS and vision providers.
  • 05Delegates coding work to Codex, OpenCode, CodeBuddy, Pi or DeepSeek Harness.
  • 06Scheduled agent tasks and an iPhone client paired over Tailscale.

Use cases

  • Describing a coding task by voice and letting a background agent build and test it.
  • Asking the assistant to watch a camera feed and notify you when a condition occurs.
  • Recalling recent work to help draft a weekly report.
  • Approving or steering agent tasks from a phone while away from your desktop or server.

Quick start

  1. 1

    Install the app globally with npm

    npm install --global nova-personal-agent@latest
  2. 2

    Open the app and configure a voice provider

    novapersonal
  3. 3

    Check that your voice pipeline keys work (optional)

    novapersonal doctor

Things to know

  • Voice, memory, camera and web search need a DashScope API key, and Tavily is optional for search.
  • Coding tasks require installing and authenticating a separate coding backend such as Codex.
  • Requires Node.js 22.13 or newer.
  • The project is at version 0.4.x and has had frequent releases and package renames, so expect changes.

Topics

#personal-agent#voice-assistant#ai-agent#agentic-ai#speech#mcp#typescript#agent#audio#audio-agent

FAQ

What is NovaPersonalAgent?
Open-source personal AI agent and voice assistant with a dual-brain design, personal memory, and delegation to coding agents. NovaPersonalAgent is a desktop and headless-server personal agent with a voice interface. A front brain keeps the conversation responsive, while a back brain runs longer tasks in the background. Users can ask Nova to clarify goals, request approval for permissions, and check results against their own criteria before taking over a task.
Is NovaPersonalAgent free to use? What license does it have?
NovaPersonalAgent is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is NovaPersonalAgent still maintained?
Actively maintained. The last commit to NovaPersonalAgent was on Oct 21, 2026.
What can I use NovaPersonalAgent for?
Describing a coding task by voice and letting a background agent build and test it. Asking the assistant to watch a camera feed and notify you when a condition occurs. Recalling recent work to help draft a weekly report. Approving or steering agent tasks from a phone while away from your desktop or server.
How do I install NovaPersonalAgent?
Install the app globally with npm: npm install --global nova-personal-agent@latest Open the app and configure a voice provider: novapersonal Check that your voice pipeline keys work (optional): novapersonal doctor
Where can I get NovaPersonalAgent?
The source code and installation instructions are on GitHub: https://github.com/deepnovacore/NovaPersonalAgent

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects