Repolode
Submit
futuregene

futuregene/future-os

A local-first AI agent with one Rust core that works across terminal UI, desktop, mobile, CLI and chat apps.

View on GitHubWebsite
Stars
111
Forks
13
Open issues
2
Last commit
2 hours ago

What is it?

FutureOS is an AI agent that runs from a single Rust backend and can be reached from a terminal UI, a desktop app, Android and iOS apps, a CLI, and IM bots. Sessions, memory and skills are shared across these surfaces, so a conversation started in one place can be continued in another.

The project focuses on controlling what the agent may do. File read, write, edit and shell tools can be governed by approval rules and sandbox tiers (off, manual, sandbox), using macOS Seatbelt, Linux Bubblewrap, or Windows restricted-token write protection depending on platform. It supports bring-your-own-key use with a built-in provider catalog and any OpenAI- or Anthropic-compatible endpoint, as well as a hosted FutureOS provider sign-in.

For long-running work, it includes a goal control plane with durable goals, todos, gates and monitors, plus scheduled tasks that run prompts on an interval or calendar schedule. Conversations can be forked, cloned and navigated as a tree.

Key features

  • 01Terminal UI, desktop, mobile, CLI and IM bot clients share one gRPC agent backend.
  • 02Configurable approval rules and OS sandbox tiers for file and shell tools.
  • 03Bring-your-own-key support with a built-in provider catalog and custom OpenAI- or Anthropic-compatible endpoints.
  • 04Durable goal tracking with verification gates for runs lasting many hours.
  • 05Scheduled recurring tasks with run history.
  • 06Forkable, cloneable session history with tree navigation.

Use cases

  • Running a long research task overnight and checking results from a phone.
  • Using one agent across a terminal workflow and a desktop app with shared memory.
  • Restricting an agent's file and shell access when working with untrusted content.
  • Connecting a self-hosted or custom LLM endpoint to a full agent interface.

Quick start

  1. 1

    Run the official install script for your platform (see README; macOS/Linux and Windows installers)

  2. 2

    Configure a model provider

    future config
  3. 3

    Launch the terminal UI

    future tui

Things to know

  • The agent needs a model provider, either your own API key or a FutureOS sign-in, before it can answer.
  • Tool sandbox availability and guarantees differ by platform, and the desktop app defaults to unrestricted tool access.

Topics

#ai-agent#llm#rust#terminal#personal-ai#sandbox#cli#desktop-app#agent-framework#ai-agents

FAQ

What is future-os?
A local-first AI agent with one Rust core that works across terminal UI, desktop, mobile, CLI and chat apps. FutureOS is an AI agent that runs from a single Rust backend and can be reached from a terminal UI, a desktop app, Android and iOS apps, a CLI, and IM bots. Sessions, memory and skills are shared across these surfaces, so a conversation started in one place can be continued in another.
Is future-os free to use? What license does it have?
future-os is open source under the MIT license. Check the license terms before using it in commercial products.
Is future-os still maintained?
Actively maintained. The last commit to future-os was on Oct 22, 2026.
What can I use future-os for?
Running a long research task overnight and checking results from a phone. Using one agent across a terminal workflow and a desktop app with shared memory. Restricting an agent's file and shell access when working with untrusted content. Connecting a self-hosted or custom LLM endpoint to a full agent interface.
How do I install future-os?
Run the official install script for your platform (see README; macOS/Linux and Windows installers) Configure a model provider: future config Launch the terminal UI: future tui
Where can I get future-os?
The source code and installation instructions are on GitHub: https://github.com/futuregene/future-os

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

Related projects