微软最近开源一个 AI 终端：Intelligent Terminal，基于 Windows Terminal 开发，在终端里内置 AI 助手。

可以自动感知命令行的输出，报错时一键把上下文丢给 AI 分析，不用手动复制粘贴，还能直接帮执行修复命令。

GitHub：http://github.com/microsoft/intelligent-terminal…

支持 GitHub Copilot、Claude Code、Codex、Gemini 等 AI Agent，会自动检测本地已安装的工具，开箱即用。

侧边栏的 AI 面板可以停靠在任意方向，复杂任务会自动开新标签页在后台跑，不打断当前工作。

所有对话数据只保存在本地，关掉会话就清除。可通过微软商店安装，和现有的 Windows Terminal 互不影响。

microsoft/intelligent-terminal

Source: https://github.com/microsoft/intelligent-terminal

Welcome to the Intelligent Terminal repo

What is Intelligent Terminal?

Intelligent Terminal is an experimental fork of Windows Terminal with native agent integration.

Intelligent Terminal works with any Agent Client Protocol (ACP)-compatible agent CLI. All you need is to install your preferred agent CLI on your PC. If you don’t have a preferred agent, we’ll get you setup with GitHub Copilot CLI.

Intelligent Terminal takes all the features you love in Windows Terminal such as: tabs, profiles, themes, settings, shells, and keyboard shortcuts, which all work the way you expect.

Read the announcement blog post for more details.

Installing and running Intelligent Terminal

Intelligent Terminal requires Windows 11 22H2 or later (22621.6060+). You also need a supported agent CLI and subscription. GitHub Copilot is the default.

Microsoft Store (recommended)

Install the Intelligent Terminal from the Microsoft Store. This allows you to always be on the latest version when we release new builds with automatic upgrades.

WinGet

winget users can download and install the latest Intelligent Terminal release by installing the Microsoft.IntelligentTerminal package:

winget install --id Microsoft.IntelligentTerminal -e

Downloads

Get Started

1. On first launch, choose your agent. Intelligent Terminal auto-detects several ACP-compatible agent CLIs on your machine (Copilot/Claude/Codex/Gemini). If none are found, it defaults to GitHub Copilot CLI and installs it for you via WinGet.
2. If you aren’t already authenticated, the agent pane walks you through sign-in.
3. Start asking questions and using the agent pane for assistance. The agent has context on your shell output, no copy-pasting needed.

If you see “running scripts is disabled on this system” or an UnauthorizedAccess error in PowerShell, your execution policy is blocking your profile and Intelligent Terminal can’t initialize shell integration. Run:

Set-ExecutionPolicy -Scope CurrentUser -ExecutionPolicy RemoteSigned

If you run into any other issues or dependency errors, see installing-dependencies.md.

Keyboard Shortcuts

All shortcuts are customizable through Intelligent Terminal settings.

Configuration

Everything is configurable through Intelligent Terminal settings, under “Agent” settings.

Features

Agent Status Bar

The agent status bar sits at the bottom of the window and gives you quick access to everything agent-related. On the left: the agent pane toggle (hotkey: Ctrl+Shift+.) and the error detection icon (hotkey: Ctrl+Alt+.), which lights up when a fixable error is detected. On the right: the agent management icon (hotkey: Ctrl+Shift+/) that opens your session management panel. It’s a persistent, minimal control surface so you’re never more than one click away from your agents.

Agent Pane

A context-aware, docked pane with your agent CLI of choice. The pane has context on your shell output across all your shells. Toggle with Ctrl+Shift+., switch focus with Ctrl+Shift+I. If the agent needs to do multiple or complex tasks, it spins up background tasks in new tabs so your active shell stays focused.

When you have multiple panes active, a small “Agent” indicator will appear on the pane that your agent has “focus” on.

Agent Management

View all active agents, their status, and past sessions. Pick up a workflow where you left off or check on a long-running task. Click the agent management icon in the status bar or press Ctrl+Shift+/ to open it.

Error Detection

When a command fails, an indicator appears in the agent status bar. Click it or press Ctrl+Alt+. to open the agent pane with the error context already loaded. The agent can explain what happened and suggest or run a fix. Configure your settings to auto-detect errors only, or to also auto-suggest fixes.

Command Palette

Type ? followed by your prompt in the Command Palette to kick off an agent task. Intelligent Terminal injects context from the active pane and starts the agent in a background tab. Use Alt+Shift+/ to jump directly into prompt mode.

Data & Privacy

Intelligent Terminal is a local transport layer. It passes your prompts and shell context to your selected agent CLI over stdio/ACP. Intelligent Terminal does not call any cloud APIs itself and does not persist conversation history, however, diagnostic logs may be written to disk and telemetry may be emitted as described below.

What data flows through Terminal

• Your prompts (what you type in the agent pane or command palette)
• Shell output context (recent command output shared with the agent for context)
• Basic environment metadata (shell type, OS version)

All of this is held in memory for the active session only and discarded when the session ends.

Where your data goes depends on your agent CLI

Terminal cannot guarantee data protections for third-party agent CLIs. When you select an agent, you’re choosing where your data goes. Review your agent vendor’s privacy policy before use. For more information on how to use GitHub Copilot responsibly, see Responsible use of GitHub Copilot.

Controls

• Choose your agent CLI at any time in Settings > Agent
• Disable auto error detection to prevent shell output from being detected automatically
• Intelligent Terminal always asks before running commands on your behalf in your shell

Intelligent Terminal only collects usage data and sends it to Microsoft to help improve our products and services. Read our privacy statement to learn more. See PRIVACY.md for details and instructions on how to disable telemetry.

Data Collection

The software may collect information about you and your use of the software and send it to Microsoft. Microsoft may use this information to provide services and improve our products and services. You may turn off the telemetry as described in the repository. There are also some features in the software that may enable you and Microsoft to collect data from users of your applications. If you use these features, you must comply with applicable law, including providing appropriate notices to users of your applications together with a copy of Microsoft’s privacy statement. Our privacy statement is located at https://go.microsoft.com/fwlink/?LinkID=824704. You can learn more about data collection and use in the help documentation and our privacy statement. Your use of the software operates as your consent to these practices.

Building the Code

Building Intelligent Terminal is the same as building Windows Terminal. See the Developer Guidance section of the Windows Terminal README for prerequisites, build instructions, and debugging steps.

Feedback

Intelligent Terminal is in an experimental stage. If you have a feature request or find a bug, submit an issue on the GitHub repository. When filing a bug, the Report a bug (collect logs) command in the Command Palette (Ctrl+Shift+P) bundles your diagnostic logs into a timestamped ZIP on your Desktop — attach it to the issue so we have full context.

Intelligent Terminal ships as a separate app and installs next to your existing Windows Terminal. If you don’t want agents in your terminal, nothing changes for you. With this model, we can learn, experiment, and iterate with you, the community, on what this evolution might look like without breaking your existing Windows Terminal flows.

Contributing

We are excited to work alongside you, our amazing community, to build and enhance Intelligent Terminal!

Before you start work on a feature/fix, please read & follow the Windows Terminal Contributor’s Guide. The contribution process is the same.

Code of Conduct

This project has adopted the Microsoft Open Source Code of Conduct. For more information, see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.

Security

If you believe you have found a security vulnerability in this repository, please report it following the instructions in SECURITY.md.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft’s Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos is subject to those third-party’s policies.