Repolode
Submit
tinyhumansai

tinyhumansai/openhuman

Open-source desktop agent app and Rust library for running many AI agents with lower token use, memory and resource cost.

Open-source alternative to:Claude CodeCodex

View on GitHubWebsite
Stars
42,162
Forks
4,164
Open issues
290
Last commit
1 hour ago

What is it?

OpenHuman is an open-source agent harness that ships as both a desktop app for end users and a Rust library for developers. It runs an agent loop with native tools, MCP servers, skills, memory that is recalled before each turn, and support for local models as well as bring-your-own-key providers.

The project focuses on cost and resource efficiency. It compresses large tool results before the model reads them, uses a small tool-search model to pick tools from a large catalog, and is built as a single compiled Rust process. The project reports benchmark results against other coding agents, which are published in a separate repository.

Developers can embed an agent in their own Rust code, or run many agents from one small server, each with its own model, tools, memory and sandbox. The project is in early beta.

Key features

  • 01Desktop app for macOS, Windows and Linux, plus a headless Rust library
  • 02Recalls relevant memory before each turn, with citations
  • 03Compresses large tool outputs before the model reads them
  • 04Supports MCP servers, skills and 119 OAuth apps via Composio
  • 05Works with local models such as Ollama and LM Studio, and BYOK providers
  • 06Sandboxed execution and an approval gate for risky actions

Use cases

  • Running a separate agent per customer on a small cloud server
  • Embedding an LLM agent with tools into a Rust application
  • Using a personal AI assistant that connects to email, calendar and messaging apps
  • Automating browser and desktop tasks with approval checkpoints

Quick start

  1. 1

    Download the desktop app from the latest release page

  2. 2

    Or install via the terminal script on macOS and Linux

    curl -fsSL https://raw.githubusercontent.com/tinyhumansai/openhuman/main/scripts/install.sh | bash
  3. 3

    Or install via PowerShell on Windows

    irm https://raw.githubusercontent.com/tinyhumansai/openhuman/main/scripts/install.ps1 | iex

Things to know

  • The project is in early beta, so expect rough edges and changes.
  • Benchmark claims come from the project's own test runs.

Topics

#ai-agents#llm#rust#desktop#mcp#local-first#privacy#agent-orchestration#ai-assistant#personal-ai

FAQ

What is openhuman?
Open-source desktop agent app and Rust library for running many AI agents with lower token use, memory and resource cost. OpenHuman is an open-source agent harness that ships as both a desktop app for end users and a Rust library for developers. It runs an agent loop with native tools, MCP servers, skills, memory that is recalled before each turn, and support for local models as well as bring-your-own-key providers.
Is openhuman free to use? What license does it have?
openhuman is open source under the GPL-3.0 license. Check the license terms before using it in commercial products.
Is openhuman still maintained?
Actively maintained. The last commit to openhuman was on Oct 19, 2026.
What can I use openhuman for?
Running a separate agent per customer on a small cloud server Embedding an LLM agent with tools into a Rust application Using a personal AI assistant that connects to email, calendar and messaging apps Automating browser and desktop tasks with approval checkpoints
How do I install openhuman?
Download the desktop app from the latest release page Or install via the terminal script on macOS and Linux: curl -fsSL https://raw.githubusercontent.com/tinyhumansai/openhuman/main/scripts/install.sh | bash Or install via PowerShell on Windows: irm https://raw.githubusercontent.com/tinyhumansai/openhuman/main/scripts/install.ps1 | iex
Where can I get openhuman?
The source code and installation instructions are on GitHub: https://github.com/tinyhumansai/openhuman

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

Related projects