Open-source AI personal assistant for planning, writing and personal knowledge, runnable on a Mac or self-hosted with your own model keys.
spytensor/openmozi
A desktop and self-hosted AI agent that runs on your machine, executes shell commands and file edits in your projects, and generates Word, Excel, PowerPoint and PDF files.
Open-source alternative to:Codex
- Stars
- 445
- Forks
- 2
- Open issues
- 0
- Last commit
- 3 months ago
What is it?
OpenMozi (MOZI) is a personal AI agent that runs on your own computer. You choose a project folder or git repository, a git branch, a permission level and an LLM, then describe a task. The agent can run shell commands, edit files, run tests, search the web, and produce documents. It checks that every file it reports as created actually exists on the filesystem before marking the task done.
The project ships as a macOS Apple Silicon desktop app built on Electron, and it can also run headless as a server with a Web UI on port 9210. Configuration is stored in a JSON file, and data is kept in local SQLite. It works with any OpenAI-compatible API and lists 27 providers, including Anthropic, OpenAI, Gemini, DeepSeek and local Ollama models.
The codebase is TypeScript on Node.js 22, using Fastify, Vercel AI SDK and React with Vite. It includes 25 bundled skills that load on demand, long-term memory across sessions, and scheduled tasks. It is heavily inspired by OpenClaw.
Key features
- 01Runs shell commands, edits files and runs tests inside a chosen project or git branch.
- 02Generates Word, PowerPoint, Excel and PDF files and previews them in the app.
- 03Keeps long-term memory across sessions.
- 04Supports scheduled and recurring tasks plus reusable task templates.
- 05Loads 25 bundled skills on demand, and custom skills via SKILL.md.
- 06Works with 27 LLM providers through OpenAI-compatible APIs, including local Ollama.
Use cases
- Letting an agent work on a local code repository while you keep control of permissions.
- Generating reports or spreadsheets from web research and local files.
- Running a self-hosted personal assistant with a Web UI on a Linux server.
- Experimenting with how an agent OS is built, since the codebase is designed to be hacked on.
Topics
FAQ
- What is openmozi?
- A desktop and self-hosted AI agent that runs on your machine, executes shell commands and file edits in your projects, and generates Word, Excel, PowerPoint and PDF files. OpenMozi (MOZI) is a personal AI agent that runs on your own computer. You choose a project folder or git repository, a git branch, a permission level and an LLM, then describe a task. The agent can run shell commands, edit files, run tests, search the web, and produce documents. It checks that every file it reports as created actually exists on the filesystem before marking the task done.
- Is openmozi free to use? What license does it have?
- openmozi is open source under the MIT license. Check the license terms before using it in commercial products.
- Is openmozi still maintained?
- Updated within 6 months. The last commit to openmozi was on Aug 7, 2026.
- What can I use openmozi for?
- Letting an agent work on a local code repository while you keep control of permissions. Generating reports or spreadsheets from web research and local files. Running a self-hosted personal assistant with a Web UI on a Linux server. Experimenting with how an agent OS is built, since the codebase is designed to be hacked on.
- Where can I get openmozi?
- The source code and installation instructions are on GitHub: https://github.com/spytensor/openmozi
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A self-hosted personal AI assistant that runs on your machine or in the cloud and connects to chat apps like DingTalk, Lark, Discord and Telegram.
Free, open-source desktop app where AI agents work on your local files, built on OpenCode and usable with any model provider or local Ollama models.
A step-by-step Go workshop that teaches you to build a coding agent on the Claude API, adding file, shell and search tools one at a time.
A TypeScript coding agent CLI where a single 'ulw' or 'mass ulw' keyword in your prompt makes it plan, delegate to multiple models, and verify the work.
Self-hosted control center for running coding agents like OpenHands, Claude Code and Codex across local, remote and cloud backends.