GitHub
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
作者提供的说明(英文原文)
Spawn, read and steer AI coding CLIs (Claude Code, Codex, Kimi and more) in local tmux windows.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"reevesagents": {
"command": "npx",
"args": [
"-y",
"reevesagents",
"mcp"
]
}
}
}在终端中运行这条命令
claude mcp add reevesagents -- npx -y reevesagents mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"reevesagents": {
"command": "npx",
"args": [
"-y",
"reevesagents",
"mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"reevesagents": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"reevesagents",
"mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月9日
注册名称已通过 GitHub 账户 mertkayacs 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(2.2.0)于昨天进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
同一分类中有:GitHub, Desktop Commander, Cloudflare MCP
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:↓ 每周下载量i★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
MCP server for terminal commands, file operations, and process management
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
ReevesAgents is free and open source under the Apache-2.0 license.
Demo · Docs · FAQ · Issues
25+ coding CLIs, one workspace.
English · Deutsch · Français · Español · Português · Italiano · Türkçe · Русский · 简体中文 · العربية
ReevesAgents is a local workspace for AI coding CLIs. It runs Claude Code, Codex, OpenCode, Hermes, Kimi, DeepSeek, Qwen, Pi, Aider, Gemini CLI, Cline CLI, Crush, Goose, Kilo Code, Letta Code, Mistral Vibe, Grok Build, Kiro CLI, Factory Droid, Qoder CLI, Warp, Junie CLI, GitHub Copilot CLI, Cursor CLI, Rovo Dev CLI, and other provider CLIs side by side in tmux. You can use it as a normal CLI/TUI/Web UI, or attach its opt-in MCP so one agent can spawn, read, steer, and stop the rest.
Provider logins stay inside each provider CLI. ReevesAgents keeps its own state as a few plain JSON files under ~/.reeves, and it runs only while you or an attached CLI are using it.
pnpm add -g reevesagents
reevesagents doctor reevesagents
Start a run from the CLI:
reevesagents spawn claude-code:lead codex:tests hermes:research \
--name "release check" \ --prompt "Review the release path, test coverage, and docs."
Open the Web UI:
reevesagents web
Let one attached agent drive the others:
reevesagents attach codex
reevesagents hosts
Restart that CLI after attach so it loads the MCP tools.
Native Windows 10 1809+ needs no WSL: reevesagents ships a built-in, tmux-compatible multiplexer (winmux) that runs panes over ConPTY. Agents keep running after you close the UI because winmux runs as a background server, just like tmux. When you attach in the TUI or CLI, the usual prefix keys work: Ctrl-b d to detach, Ctrl-b n / p to move between agent windows, Ctrl-b 0-9 to jump to a window.
| Surface | Use it for |
|---|---|
| TUI | Keyboard-first run control inside the terminal. |
| Web UI | Local visual view of runs, panes, agents, approvals, and history. |
| CLI | Scripts, smoke checks, spawning, state cleanup, and tmux jumps. |
| Agent Control MCP | One trusted CLI can spawn and drive other CLIs through local tools. |
| tmux | Real provider CLI windows that keep running after the UI closes. |
ReevesAgents is local by design. State is plain JSON under ~/.reeves, and the CLIs it starts are the same CLIs you already use by hand.
Four tools sit next to the coding CLIs. Perplexity CLI searches the web, Google Colab CLI runs scripts and notebooks on CPU, GPU and TPU runtimes, Hugging Face CLI downloads and uploads models and datasets and runs jobs on the Hub, and Ollama runs local or cloud models from the terminal. Agents call them from their own shell.
reevesagents tools lists the four with a purpose, an example command, and whether each is installed and signed in (for Ollama: reachable). Add --json for scripts. An agent with the MCP attached gets the same list from list_tools.
Install
ReevesAgents needs Node.js 22.12+, tmux 3.0+, and at least one supported provider CLI installed and authenticated. It supports macOS, Linux, and WSL.
# Homebrew
brew tap mertkayacs/reevesagents brew install reevesagents
pnpm add -g reevesagents
npm install -g reevesagents
pnpm dlx reevesagents doctor npx reevesagents doctor
Pin a version by replacing :
pnpm add -g reevesagents@<version>
Source install:
git clone https://github.com/mertkayacs/reevesagents.git
cd reevesagents pnpm install pnpm build pnpm link --global reevesagents doctor
Screenshots
The TUI and Web UI drive the same local runs:
Commands
No arguments launches the TUI.
| Command | Purpose | ||||
|---|---|---|---|---|---|
reevesagents | Launch the TUI. | ||||
spawn [spec...] | Start a run. Each spec is provider[:nickname[:model]]. | ||||
add [spec...] | Add agents to the most recent active run. | ||||
runs | List active runs. | ||||
agents [run-id] | List agents across runs or inside one run. | ||||
open | Jump to a run or agent tmux window. | ||||
peek | Print recent output from one agent. | ||||
send | Paste text into an agent without submitting. | ||||
key | Send enter, escape, arrows, tab, space, backspace, or ctrl-c. | ||||
interrupt | Send Ctrl-C to one agent. | ||||
stop | Stop a run. Requires --yes or ALLOW_DESTRUCTIVE=1. | ||||
kill | Stop one agent. Requires --yes or ALLOW_DESTRUCTIVE=1. | ||||
setup | First-run check. --attach connects every installed host CLI. | ||||
doctor | Check Node, tmux, state, and provider CLIs. | ||||
web | Start the loopback-only Web UI. | ||||
providers | List provider ids, aliases, models, and availability. | ||||
tools | List Perplexity, Google Colab, Hugging Face and Ollama with purpose, example, and install and sign-in state. | ||||
approvals | List pending approval requests. | ||||
approve / deny | Resolve one approval request. | ||||
hosts | Show which host CLIs have ReevesAgents attached. | ||||
attach [cli] | Connect the Agent Control MCP to one host CLI, or all installed hosts. | ||||
detach | Remove that MCP connection from one host CLI. | ||||
skills [action] | Install, remove, or inspect the ReevesAgents skill. | ||||
mcp | Start the MCP server over stdio. Host CLIs run this. | ||||
config [key] [value] | Show or update editable settings. | ||||
flow start | Attach a dev → review → fix workflow to existing agents in a run. | ||||
flow status | Show progress, agent questions, open incidents, and the last tick receipt. | ||||
flow tick / flow tick-all | Advance one flow or all flows; --dry-run observes without sending. Automatic progression requires an external scheduler. | ||||
| `flow plan import\ | attach\ | next\ | status\ | set` | Import a plan, attach briefs, inspect the next item or progress, and make explicit owner decisions. |
flow add-spec / flow drop-spec | Manage standalone queued work. | ||||
| `flow pause\ | resume\ | stop\ | release-round` | Control the workflow or release an eligible unreviewed round. | |
flow logs / flow notes | Inspect transitions and review findings. | ||||
| `flow incidents list\ | claim\ | ack\ | resolve` | Inspect incidents, claim or acknowledge ownership, and resolve with a note. | |
notifications list / notifications read | Inspect notification history and mark events read. | ||||
presets | List saved run presets. | ||||
save-preset | Save a live run as a preset. | ||||
start-preset | Start a run from a preset. | ||||
delete-preset | Delete a preset. | ||||
delete | Delete one ended agent record. Requires confirmation. | ||||
delete-run | Delete one ended run and archive it. Requires confirmation. | ||||
history | List archived runs. | ||||
delete-history | Delete one archived history record. Requires confirmation. | ||||
reap | End zombie agents and agents past max_lifetime_ms, and kill orphan tmux sessions no run record owns. |
Common flags:
--json: available on script-facing list and action commands.--name : name a run.--cwd : run agents from a directory.--prompt : paste startup text into each spawned agent.--skip: skip provider permission prompts for unattended workers.--run : add agents to a specific run.--port and --no-open: Web UI startup options.Agent Control
Agent Control is an optional MCP server. Attach it only to a CLI you trust to drive local tools:
reevesagents attach claude
reevesagents hosts
After restart, that CLI receives tools to spawn, read, send_text, send_key, interrupt, kill, stop, manage approvals, manage presets, and inspect hosts. The provider catalog is also exposed as reevesagents://providers.
In Claude Code you can install the connection and its operating instructions together as a plugin:
claude plugin marketplace add mertkayacs/reevesagents
claude plugin install reevesagents@reevesagents
The server is listed in the official MCP Registry as io.github.mertkayacs/reevesagents.
<img src="https://raw.githubusercontent.com/mertkayacs/reevesagents/master/docs/assets/reevesagents-steer.webp" width="560" alt="A gold-framed lead pane sending messages along gold thread
只显示了开头部分;其余内容请在仓库中阅读