DEV.co
AI Frameworks · microsoft

UFO

UFO is a Python-based desktop automation framework by Microsoft that evolved into UFO³, supporting both single-device Windows automation (UFO²) and multi-device orchestration (Galaxy). It uses LLMs to understand GUI elements and execute tasks across Windows, Linux, and Android platforms through a DAG-based task decomposition model.

Source: GitHub — github.com/microsoft/UFO
9.3k
GitHub stars
1k
Forks
Python
Primary language
MIT
License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
Repositorymicrosoft/UFO
Ownermicrosoft
Primary languagePython
LicenseMIT — OSI-approved
Stars9.3k
Forks1k
Open issues75
Latest release3.0.7 (2026-06-12)
Last updated2026-06-26
Sourcehttps://github.com/microsoft/UFO

What UFO is

UFO³ comprises two components: UFO² (a Windows AgentOS with UIA/Win32/COM integration, hybrid GUI+API actions, and speculative multi-action optimization) and Galaxy (a multi-device orchestration framework with ConstellationAgent DAG planning, TaskOrchestrator async execution, capability-based device assignment, and a WebSocket-based Unified Agent Interaction Protocol). Built on Python 3.10+, it integrates MCP for tool augmentation.

Quickstart

Get the UFO source

Clone the repository and explore it locally.

terminalbash
git clone https://github.com/microsoft/UFO.gitcd UFO# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Cross-Device Workflow Automation

Orchestrate complex tasks spanning Windows, Linux, and Android devices—e.g., data ingestion on one platform, processing on another, reporting on a third—using Galaxy's DAG-based task decomposition and asynchronous parallel execution.

Windows Desktop Automation & RPA

Use UFO² standalone for single-machine GUI automation with deep Windows integration (UIA, Win32, COM), hybrid actions, and speculative multi-action to reduce LLM calls by ~51% for proven desktop workflows.

Multi-App Enterprise Task Automation

Automate complex sequences across multiple Windows applications (e.g., CRM to spreadsheet to email) with UFO²'s HostAgent + AppAgent architecture, leveraging visual + UIA detection for robust element identification.

Implementation considerations

  • Python 3.10+ and pip/package management required; test in staging before production desktop rollout to validate UIA/Win32 compatibility with your application stack.
  • UFO² is LTS (long-term support) and stable; UFO³ Galaxy is in active development (last push 2026-06-26). Start with UFO² for near-term needs, prototype Galaxy for future multi-device scenarios.
  • LLM integration (model choice, token costs, latency) directly impacts performance; speculative multi-action can reduce calls but requires tuning for your workflow patterns.
  • Multi-device Galaxy deployments require network connectivity, WebSocket support, and agent discovery/registration; plan for monitoring and fault-tolerance (automatic reconnection noted).
  • MCP (Model Context Protocol) integration implies vendor lock-in to MCP-compatible tools; review available tool ecosystems for your use case before committing to device agent extensions.

When to avoid it — and what to weigh

  • Require Non-Windows Production Support Today — UFO² is battle-tested on Windows only; Galaxy's Android/Linux support is nascent (November 2025 release). If you need production-grade cross-platform agents now, plan for Galaxy to still be in active development.
  • Simple, Low-Code/No-Code Automation — UFO's strength is LLM-driven intelligent automation. For purely rule-based clicks or straightforward RPA, simpler tools (UIPath, Blue Prism) may be faster to configure and have broader enterprise support.
  • Strict Commercial Licensing Requirements — While MIT-licensed, UFO is open-source and backed by Microsoft R&D. If you require explicit vendor support contracts, professional SLAs, or pre-negotiated indemnification for proprietary deployment, evaluate commercial terms first.
  • Isolated Environments Without Internet/GPU — UFO requires LLM calls (presumably cloud-based). Air-gapped or fully offline automation will require local LLM integration; Galaxy's WebSocket AIP is designed for networked multi-device setups, not isolated networks.

License & commercial use

UFO is released under the MIT License, a permissive OSI-approved license permitting commercial use, modification, and distribution with minimal restrictions (retain copyright notice and license text).

MIT license explicitly permits commercial use without vendor consent or royalty. However, UFO is open-source, community-maintained research software. No commercial support, SLA, or indemnification is stated; for enterprise production use, assess internal risk tolerance or engage Microsoft directly for any required support/licensing agreement.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationStrong
License clarityClear
Deployment complexityModerate
DEV.co fitStrong
Assessment confidenceHigh
Security considerations

No security audit or vulnerability disclosure policy stated. UFO automates GUI/API interactions on Windows and other platforms; ensure isolated test environments before production use. WebSocket-based AIP (Galaxy) should use TLS in production; validate encryption in your deployment. Sensitive credentials in task definitions require secure handling (no details provided). Community-maintained, so security response time is unknown; monitor GitHub issues and upstream LLM provider updates.

Alternatives to consider

UIPath / Blue Prism

Mature, enterprise-grade RPA platforms with professional support, SLAs, and pre-built process templates. Best if you need vendor accountability and low-code workflows, but higher licensing cost.

Anthropic Codeium / GitHub Copilot (Agents mode)

Cloud-native LLM agents with multi-step reasoning and tool-use. Simpler setup for lightweight automation but less control over device/GUI integration compared to UFO's native Windows APIs.

n8n / Make (Integromat)

Visual workflow automation platforms with broad SaaS/API integrations. Ideal for cloud-native, API-first processes; lacks deep desktop GUI automation capabilities but easier for non-technical users.

Software development agency

Build on UFO with DEV.co software developers

UFO combines proven Windows automation (UFO²) with emerging multi-device orchestration (Galaxy). Evaluate your use case, test in staging, and contact our team to plan integration with your automation roadmap.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

Related on DEV.co

Explore the category and the services that help you build with it.

UFO FAQ

Can I use UFO for production automation on non-Windows platforms today?
UFO² is production-ready on Windows only. UFO³ Galaxy announced support for Linux and Android in November 2025 (latest release 3.0.7, June 2026), but it is in active development. Plan for testing and potential workarounds; Windows via UFO² remains the battle-tested path.
Is UFO free to use commercially?
Yes, MIT license permits commercial use without royalty. However, no commercial support, SLA, or indemnification is provided. For enterprise production, assess internal risk tolerance or contact Microsoft for any required support agreements.
What's the difference between UFO² and UFO³ Galaxy?
UFO² is a single-device Windows AgentOS (LTS, stable); UFO³ Galaxy adds multi-device orchestration (active development). UFO² can serve as a Galaxy device agent. Start with UFO² for near-term single-machine automation; adopt Galaxy for future cross-device workflows.
How much does it cost to integrate an LLM (e.g., GPT-4, Claude)?
UFO is free; you pay for LLM API usage separately (OpenAI, Anthropic, etc.). Speculative multi-action can reduce calls ~51%, lowering token cost. No hidden fees from UFO itself.

Work with a software development agency

DEV.co is a software development agency delivering custom software development services to companies building on open source. Our software developers and web developers design, integrate, and ship production systems — spanning web development, APIs, AI, data, and cloud. If UFO is part of your ai frameworks roadmap, our team can implement, customize, migrate, and maintain it.

Ready to Automate Complex Desktop Workflows?

UFO combines proven Windows automation (UFO²) with emerging multi-device orchestration (Galaxy). Evaluate your use case, test in staging, and contact our team to plan integration with your automation roadmap.