能力标签

🔌 MCP 🤖 Agent 🔄 工作流 🐳 Docker 💻 CLI 🧠 Claude ✨ GPT

🔌

MCP工具

终端AI代理

基于 Rust · 让 AI 助手直接操作你的系统与工具

英文名：unterm

⭐ 6 Stars 💻 Rust 📄 NOASSERTION 🏷 AI 8.0分

8.0AI 综合评分

mcpagent-toolsclaudecodex

⚙️ 配置说明 📺 TG 频道

✦ AI Skill Hub 推荐

经 AI Skill Hub 精选评估，终端AI代理获评「强烈推荐」。这款MCP工具在功能完整性、社区活跃度和易用性方面表现出色，AI 评分 8.0 分，适合有一定技术背景的用户使用。

📚 深度解析

终端AI代理是一款基于 MCP（Model Context Protocol）标准协议的 AI 工具扩展。MCP 协议由 Anthropic 开发并开源，旨在建立 AI 模型与外部工具之间的标准化通信接口，目前已被 Claude Desktop、Claude Code、Cursor 等主流 AI 工具采纳。

通过安装终端AI代理，你的 AI 助手将获得额外的工具调用能力，可以用自然语言直接操控该工具的功能，无需学习复杂的命令行语法。MCP 工具的核心价值在于"一次配置，永久增强"——配置完成后，每次与 AI 对话时都可以无缝调用这些工具。

在技术实现上，MCP 工具通过标准的 JSON-RPC 协议与 AI 客户端通信，工具的功能以"工具列表"的形式暴露给 AI 模型，AI 可以按需调用。终端AI代理提供了结构化的工具调用接口，使 AI 模型能够精确地理解和使用每个功能点，显著降低 AI 在工具使用上的错误率。

与传统的 API 集成相比，MCP 工具的优势在于无需编写代码——用户只需在配置文件中添加几行 JSON，即可让 AI 获得全新能力。AI Skill Hub 将终端AI代理评为 AI 评分 8.0 分，属于同类工具中的优质选择。

📋 工具概览

终端AI代理是一款遵循 MCP（Model Context Protocol）标准协议的 AI 工具扩展。通过 MCP 协议，它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务，实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用，都可以通过自然语言在 AI 对话中直接触发，极大提升生产效率。

GitHub Stars

⭐ 6

开发语言

Rust

支持平台

Windows / macOS / Linux

维护状态

轻量级项目，按需更新

开源协议

NOASSERTION

AI 综合评分

8.0 分

工具类型

MCP工具

Forks

—

📖 中文文档

以下内容由 AI Skill Hub 根据项目信息自动整理，如需查看完整原始文档请访问底部「原始来源」。

📌 核心特色

通过标准 MCP 协议与 Claude、Cursor 等主流 AI 客户端深度集成
提供结构化工具调用接口，显著降低 AI 集成复杂度
支持 Claude Desktop 和 Claude Code 无缝接入，开箱即用
可与其他 MCP 工具组合叠加，构建完整 AI 工作站
轻量无侵入设计，不影响现有系统架构

🎯 主要使用场景

在 Claude Desktop 对话中直接调用本地工具，实现 AI 与系统的深度联动
通过自然语言驱动复杂的多步骤自动化任务，代替繁琐手动操作
将多个 MCP 工具组合使用，构建个人专属 AI 工作站

以下安装命令基于项目开发语言和类型自动生成，实际以官方 README 为准。

安装命令

# 方式一：通过 Claude Code CLI 一键安装
claude skill install https://github.com/zhitongblog/unterm

# 方式二：手动配置 claude_desktop_config.json
{
  "mcpServers": {
    "--ai--": {
      "command": "npx",
      "args": ["-y", "unterm"]
    }
  }
}

# 配置文件位置
# macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
# Windows: %APPDATA%/Claude/claude_desktop_config.json

📋 安装步骤说明

确认已安装 Node.js（v18 或以上版本）
打开 Claude Desktop 或 Claude Code 的 MCP 配置文件
按「交给 Agent 安装 → Claude Desktop」标签中的 JSON 配置填入 mcpServers 字段
保存配置文件并重启 Claude 客户端
重启后，在对话中即可使用本工具

以下用法示例由 AI Skill Hub 整理，涵盖最常见的使用场景。

常用命令 / 代码示例

# 安装后在 Claude 对话中直接使用
# 示例：
用户: 请帮我用 终端AI代理 执行以下任务...
Claude: [自动调用 终端AI代理 MCP 工具处理请求]

# 查看可用工具列表
# 在 Claude 中输入："列出所有可用的 MCP 工具"

以下配置示例基于典型使用场景生成，具体参数请参照官方文档调整。

配置示例

// claude_desktop_config.json 配置示例
{
  "mcpServers": {
    "__ai__": {
      "command": "npx",
      "args": ["-y", "unterm"],
      "env": {
        // "API_KEY": "your-api-key-here"
      }
    }
  }
}

// 保存后重启 Claude Desktop 生效

📑 README 深度解析真实文档完整度 55/100 查看 GitHub 原文 →

以下内容由系统直接从 GitHub README 解析整理，保留代码块、表格与列表结构。

Unterm

The terminal AI agents can drive.

Cross-platform terminal (macOS / Linux / Windows) built on a customized WezTerm engine, with one design bet: the terminal itself is controllable from the outside by any AI agent over MCP. Claude Code, Cursor, Aider, Continue, your own scripts — they all get the same JSON-RPC surface to spawn shells, run commands, read pane state, capture screenshots, change settings, and record sessions.

The other 2026 terminals each pick a different side: Warp embeds AI inside a closed cloud (Oz), Ghostty stays out of your way and lets you bring your own tools, iTerm2 is Mac-only. Unterm picks the third side — terminal as MCP-controllable surface, deliberately keep AI features out of the terminal, let external agents grip it through the API.

Practical implications:

Every Unterm window starts a local MCP server (line-delimited JSON-RPC over TCP) and a local HTTP settings server (Web Settings page) on auto-allocated ports. Both are auth-token gated, both are 127.0.0.1-only, no cloud round trip.
Settings live in the browser, not the terminal. Cell-grid TUIs can't deliver modern form UX (no proper text inputs, no live preview, no color picker). The in-terminal ▼ overlay is intentionally minimal — six quick actions and a link to the Web Settings page.
9 languages out of the box: en / 简体中文 / 繁體中文 / 日本語 / 한국어 / Deutsch / Français / Italiano / हिन्दी. Auto-detects from system locale, can be overridden in Web Settings or via unterm-cli lang set <code>.
Multi-instance discovery: every running Unterm process owns one NATO-named instance (alpha, bravo, charlie…) and writes its ports + auth token to ~/.unterm/instances/<name>.json. Agents that drive several windows at once enumerate that directory.
Cross-platform parity is a correctness property: if a feature works on Windows but bails on macOS or Linux, that's a bug, not "not supported yet."
Subtraction over decoration: no AI overlay inside the terminal, no inline image render that wedges the GUI, no in-terminal custom right-click chrome, no Cmd+Q confirmation, no manual proxy URL config (auto-detected from system). Finder integration on macOS uses the native Finder right-click extension and Services.

Built on top of the WezTerm engine for renderer / font / TUI / SSH / mux work, with a thin product layer on top.

---

What's new

v0.12 — synchronous pre-show paint kills the long-standing Windows white flash on launch. The frame is rendered before the window is shown, not after.
v0.9 — multi-instance discovery with NATO-phonetic names. Each running Unterm writes ~/.unterm/instances/<name>.json; ~/.unterm/active.json points at the current foreground instance for single-target agents.
v0.7 — Windows defaults to UTF-8 out of the box. PowerShell and cmd.exe spawns are wrapped to set the code page; no more mojibake on Chinese filenames.
v0.5 — dogfood milestone. Default window sizing tuned for real use, scrollback line count is configurable via ~/.unterm/scrollback.json, TERM_PROGRAM overridable via ~/.unterm/compat.json, background update poller writes ~/.unterm/update_check.json.

---

Features

GPU-accelerated rendering on all three platforms (Metal / OpenGL / DirectX via ANGLE).
MCP server on 127.0.0.1:<auto-port> (default 19876) — JSON-RPC over TCP, auth-token gated. Method namespaces: session, exec, screen, signal, orchestrate, proxy, workspace, capture, policy, system, server, instance, profile.
Web Settings UI on 127.0.0.1:<auto-port> (default 19877) — open in any browser via unterm-cli settings open or the Settings (Web) item in the ▼ menu. Tailwind-styled SPA, supports all 9 languages, keyboard + mouse.
Auto proxy detection — reads macOS System Preferences / Windows registry / GNOME gsettings / $HTTPS_PROXY, falls back to scanning common local ports. The single proxy.json toggle is {"enabled": true|false} — no manual URL configuration needed.
Region screenshots from the status bar (left-click excludes the Unterm window, right-click includes it). PNG lands on disk under ~/.unterm/screenshots/, on the system image clipboard, and the path on the text clipboard.
Scrolling (long) screenshots, both directions: capture.scrollback re-renders a pane's entire history into one tall PNG headlessly (exact fonts/theme, streaming-encoded, works while occluded); capture.window_scroll long-shots another app's window by synthesizing wheel events and stitching frames via row-hash matching with sticky-header/footer detection (macOS). Both also in the ▼ menu and unterm-cli screenshot --scrollback / --scroll-app.
Session recording → markdown with OSC 133 block segmentation and built-in redaction (GitHub tokens / KEY=value / 40+ char hex/base64 patterns are masked). Recordings are stored in the project directory under <cwd>/.unterm/sessions/<date>/<tab>-<time>.md, or in ~/.unterm/sessions/_orphan/ when no writable project context.
Right-click is a direct gesture, not a menu: with a selection it copies and clears; without selection it pastes.
Slim quick-action overlay on the tab bar's ▼ button:
Change Working Directory (cd current pane)
Open Folder in New Tab
Split Right (left/right pane split)
Toggle Session Recording
Export Current Session
Settings (Web)
macOS-native window decorations (traffic-light buttons + native title bar); Windows uses Windows Terminal-style integrated title buttons; Linux uses client-side decorations.

---

Windows — MSI (requires WiX 6 at .\.tools\wix.exe — install via `dotnet tool install --tool-path .\.tools wix --version 6.0.1`)

pwsh -File ci/build-msi.ps1


macOS code-signing + notarization is **local-only** (no CI step) so the
Developer ID `.p12` private key never has to leave your Mac. One-time
setup, on the Mac that holds the cert:

bash xcrun notarytool store-credentials UntermNotary \ --apple-id <your-apple-id> --team-id 6NQM3XP5RF ```

Install

Pre-built artifacts are published on GitHub Releases:

https://github.com/zhitongblog/unterm/releases

Platform	Artifact
macOS	`Unterm-macos-<version>.dmg` (universal arm64+x86_64, signed + notarized)
Linux	`unterm-<version>.deb` or `Unterm-<version>-x86_64.AppImage`
Windows	`Unterm-<version>-x64.msi` or `Unterm-windows-<version>.zip`

Screenshots

unterm-cli screenshot [--include-window] [-o FILE]

Long screenshot of a pane's ENTIRE scrollback (headless re-render -> tall PNG)

unterm-cli screenshot --scrollback [--pane N] [--max-rows N] [-o FILE]

Long screenshot of ANOTHER app's window: scroll + stitch (macOS)

unterm-cli screenshot --scroll-app Safari [--scroll-title SUBSTR] [--max-frames N] [-o FILE] ```

Pass --json to any subcommand for raw JSON-RPC output (suitable for scripts). Pass --lang <code> to override the locale for one invocation. Pass --instance <id> (or set UNTERM_INSTANCE=<id>) when several Unterm windows are open and you need a deterministic target.

Multi-instance discovery is exposed over MCP rather than as a separate CLI list command — call instance.list against any running Unterm's MCP port, or just ls ~/.unterm/instances/.

Settings + Web UI

unterm-cli settings open # open the Web Settings page unterm-cli theme list / set <id> # standard / midnight / daylight / classic / notion-dark / notion-light unterm-cli lang list / set <code> / current # en-US / zh-CN / zh-TW / ja-JP / ko-KR / de-DE / fr-FR / it-IT / hi-IN

Configuration

User config lives at:

Platform	Location
macOS	`~/.unterm/`
Linux	`~/.unterm/`
Windows	`%USERPROFILE%\.unterm\`

Files:

File	Purpose
`server.json`	Active instance's MCP/HTTP ports + auth token + pid (auto, mirrors the active instance for back-compat)
`active.json`	Pointer at the current foreground instance id (auto, updated only when previous active dies)
`instances/<name>.json`	Per-instance metadata (NATO id, ports, token, pid, started_at, version, platform)
`auth_token`	Legacy mirror of the active auth token (for back-compat)
`proxy.json`	`{"enabled": true\|false}` — URLs auto-detected
`theme.json`	Active theme id
`lang.json`	Persisted locale override
`compat.json`	`{"term_program": "..."}` override for `$TERM_PROGRAM`
`scrollback.json`	Override the default scrollback line count
`update_check.json`	Background update-poller state (last check, latest seen version)
`onboarded.json`	First-run flags (which `▼` items have been seen)
`recording.json`	Recording config (redaction patterns, etc.)
`sessions/`	Recording metadata index (per-project subdirs)
`screenshots/`	Region screenshots (PNG)