OpenRig Upgrade Procedure
mvschwarz/openrig
Walks an agent through upgrading the OpenRig CLI and daemon one observed step at a time, keeping live seats alive and reconciling managed plugin files.
Operates tmux sessions like an administrator: reads pane output, sends keys, inspects many panes at once, and coordinates multiple AI terminals through a swarm state script, built on oh-my-tmux.
Install Auto tmux Operator
$ npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install tradecatlabs/vibe-coding-cn auto-tmux --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/tradecatlabs/vibe-coding-cn.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/auto-tmux .claude/skills/auto-tmux && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "auto-tmux" agent skill from https://github.com/tradecatlabs/vibe-coding-cn/tree/develop/skills/auto-tmux into .claude/skills/auto-tmux/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "auto-tmux", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/tradecatlabs/vibe-coding-cn/tree/develop/skills/auto-tmuxType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install tradecatlabs/vibe-coding-cn auto-tmux --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/tradecatlabs/vibe-coding-cn.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/auto-tmux .agents/skills/auto-tmux && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "auto-tmux" agent skill from https://github.com/tradecatlabs/vibe-coding-cn/tree/develop/skills/auto-tmux into .agents/skills/auto-tmux/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "auto-tmux", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install tradecatlabs/vibe-coding-cn auto-tmux --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/tradecatlabs/vibe-coding-cn.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/auto-tmux .cursor/skills/auto-tmux && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "auto-tmux" agent skill from https://github.com/tradecatlabs/vibe-coding-cn/tree/develop/skills/auto-tmux into .cursor/skills/auto-tmux/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "auto-tmux", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/tradecatlabs/vibe-coding-cn.git --path skills/auto-tmux--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install tradecatlabs/vibe-coding-cn auto-tmux --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/tradecatlabs/vibe-coding-cn.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/auto-tmux .gemini/skills/auto-tmux && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "auto-tmux" agent skill from https://github.com/tradecatlabs/vibe-coding-cn/tree/develop/skills/auto-tmux into .gemini/skills/auto-tmux/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "auto-tmux", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install tradecatlabs/vibe-coding-cn auto-tmuxInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/tradecatlabs/vibe-coding-cn.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/auto-tmux .github/skills/auto-tmux && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "auto-tmux" agent skill from https://github.com/tradecatlabs/vibe-coding-cn/tree/develop/skills/auto-tmux into .github/skills/auto-tmux/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "auto-tmux", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install tradecatlabs/vibe-coding-cn auto-tmux --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/tradecatlabs/vibe-coding-cn.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/auto-tmux .opencode/skills/auto-tmux && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "auto-tmux" agent skill from https://github.com/tradecatlabs/vibe-coding-cn/tree/develop/skills/auto-tmux into .opencode/skills/auto-tmux/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "auto-tmux", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
auto-tmux- GitHub stars
- 17k
- Token cost
- ~4.7k tokens
- SKILL.md length
- 751 words
- Files
- 58 (incl. scripts, references, assets)
- Skills in repo
- 17
- Repo updated
- First seen
- Licence
- MIT
At a glance
Operates tmux sessions like an administrator: reads pane output, sends keys, inspects many panes at once, and coordinates multiple AI terminals through a swarm state script, built on oh-my-tmux.
- Works in 3 steps: skills/auto-tmux/scripts/auto-tmux.sh… → skills/auto-tmux/scripts/auto-tmux.sh… → skills/auto-tmux/scripts/auto-tmux.sh…
- Reading recent output from one or many tmux panes at once
- SKILL.md covers When to Use This Skill, Not For / Boundaries, Quick Reference and Rules & Constraints, plus 6 more sections
- Calls bash, git and codex
What it does
The skill prefers a bundled auto-tmux.sh wrapper over raw tmux commands for reading a pane's recent output, sending keystrokes, batch-inspecting several panes, and rescuing or recording a stuck terminal, falling back to native tmux commands such as capture-pane and send-keys only when needed. A separate swarm-state.sh script manages a task queue, locks, status and results for coordinating multiple AI terminals, and the skill ships a vendored copy of the oh-my-tmux configuration under assets so its double-prefix shortcuts and pane sync commands are available without touching the user's own tmux config.
Stated boundaries: it does not create a session, window or pane that does not already exist, it never edits ~/.tmux.conf directly (changes go in .tmux.conf.local per oh-my-tmux convention), it will not blindly send destructive keys such as kill-server without first capturing the pane to confirm context, and it treats scripts as a convenience, not a way around requiring human confirmation for sensitive sessions, password entry or production database work. It also calls for upgrading tmux first if the installed version is older than 2.6.
When your agent uses it
- Reading recent output from one or many tmux panes at once
- Sending confirmation keystrokes or commands to a specific pane safely
- Coordinating several AI agent terminals through a shared task queue
- Rescuing a stuck tmux pane without guessing at its current state
Example prompts
- “Capture the last 120 lines from the build pane and tell me if it finished.”
- “Send make test to the worker pane and press Enter.”
- “Set up a swarm task queue across the three AI terminals in this session.”
Requirements
- tmux 2.6 or later
- Bash
Workflow steps
3 steps, taken from the first numbered list in SKILL.md.
- skills/auto-tmux/scripts/auto-tmux.sh topology 找到目标 :.。
- skills/auto-tmux/scripts/auto-tmux.sh rescue -t --pattern "(y/n)" --reply y --dry-run 确认命中。
- skills/auto-tmux/scripts/auto-tmux.sh rescue -t --pattern "(y/n)" --reply y 发送确认。
What it can do on your machine
Read from SKILL.md and the folder at commit 5b76a8f. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Ships 1 file in scripts/, which the agent can run.
Shell commands in SKILL.md call:
bashgitcodex
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use git, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Auto tmux Operator loads about 4.7k tokens when it runs, and up to ~25k if it reads all its reference files. Until then it costs about 49 tokens; SKILL.md has 751 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.
SKILL.md
The full file from tradecatlabs/vibe-coding-cn at commit 5b76a8f, republished under its MIT licence (© tradecatlabs). 751 words, ~4,657 tokens.
Download SKILL.mdSave it as
.claude/skills/auto-tmux/SKILL.md (or your agent's skills folder). This skill also uses 57 other files; get the full folder from GitHub.- name
- auto-tmux
- description
- tmux 自动化操控：用 scripts/auto-tmux.sh 安全读取、发送、巡检、救援、录制 session|window|pane，用 swarm-state.sh 管理蜂群任务/锁/状态，并基于 oh-my-tmux 组织多 AI 终端协作。触发：capture-pane、send-keys、批量巡检、蜂群 AI 协作、卡死救援、tmux 工作台初始化。
auto-tmux Skill
让 AI 像熟练运维一样操作 tmux：优先使用 scripts/auto-tmux.sh 安全封装读取终端输出、发送按键、批量巡检、协作/救援其他终端；使用 scripts/swarm-state.sh 管理蜂群任务、锁、状态和结果；必要时再回退到 tmux 原生命令。默认兼容 tools/external/.tmux（gpakosz/oh-my-tmux），并在本技能内通过 assets/oh-my-tmux 与 assets/tmux-src 暴露上游 submodule。
When to Use This Skill
触发条件（满足其一即可）：
- 需要远程读取/复制某个 tmux pane 的最新输出（日志、提示、错误）。
- 需要向指定 pane 发送按键/命令（确认
y、Enter、Ctrl+C、广播同一窗口）。 - 需要批量巡检/接管多 AI 终端（蜂群协作、自动救援卡死任务）。
- 用户显式要求进入 Codex Pilot Mode：让当前主 Codex 通过 tmux 控制一个持久化、交互式、对人类可见的 worker Codex pane。
- 需要初始化 AI 多终端工作台、等待输出 pattern、录制 pane 日志或保存巡检证据。
- 需要给 tmux 蜂群建立任务队列、状态日志、文件锁和结果报告。
- 需要快速回忆 oh-my-tmux 快捷键、前缀或同步面板操作。
- 需要在当前仓库复用
tools/external/.tmux配置并避免修改主配置。 - 需要从技能目录内快速查看 oh-my-tmux 配置或 tmux 上游源码入口。
Not For / Boundaries
- 不用于执行非 tmux 环境（目标不存在时先创建 session/window/pane）。
- 不直接编辑
~/.tmux.conf/.tmux.conf（遵循 oh-my-tmux：修改请放到.tmux.conf.local）。 - 不盲发破坏性按键：
kill-server、rm -rf等必须先 capture-pane 确认目标上下文。 - 不把脚本当成权限绕过器；敏感会话、密码输入、生产数据库操作仍需人工确认。
- 如果 tmux 版本 < 2.6，需先升级再使用本技能的命令组合。
Quick Reference
常用模式（≤20 条，直接复制即用）
脚本帮助与语法检查
bash -n skills/auto-tmux/scripts/auto-tmux.sh
bash -n skills/auto-tmux/scripts/swarm-state.sh
bash -n skills/auto-tmux/scripts/swarm-brief.sh
bash -n skills/auto-tmux/scripts/swarm-watch.sh
bash -n skills/auto-tmux/scripts/swarm-archive.sh
bash -n skills/auto-tmux/scripts/swarm-board.sh
bash -n skills/auto-tmux/scripts/swarm-deps-graph.sh
bash -n skills/auto-tmux/scripts/swarm-export.sh
bash -n skills/auto-tmux/scripts/swarm-timeline.sh
bash -n skills/auto-tmux/scripts/swarm-blockers.sh
bash -n skills/auto-tmux/scripts/swarm-results.sh
bash -n skills/auto-tmux/scripts/swarm-report-pack.sh
bash -n skills/auto-tmux/scripts/swarm-assign.sh
bash -n skills/auto-tmux/scripts/swarm-health.sh
bash -n skills/auto-tmux/scripts/remote-readonly.sh
bash -n skills/auto-tmux/scripts/record-summary.sh
bash -n skills/auto-tmux/scripts/check-jsonl.sh
bash -n skills/auto-tmux/scripts/review-checklist.sh
bash -n skills/auto-tmux/scripts/verify-report-pack.sh
bash -n skills/auto-tmux/scripts/incident-report.sh
bash -n skills/auto-tmux/scripts/completion.bash
bash -n skills/auto-tmux/scripts/safety-check.sh
bash -n skills/auto-tmux/scripts/render-swarm-prompt.sh
bash -n skills/auto-tmux/scripts/swarm-dispatch.sh
bash -n skills/auto-tmux/scripts/validate-auto-tmux.sh
skills/auto-tmux/scripts/auto-tmux.sh help
skills/auto-tmux/scripts/swarm-state.sh help查看快捷键帮助（oh-my-tmux 双前缀：主 C-a，备用 C-b，前缀后按 ?）
[按] <prefix> ?枚举拓扑（脚本优先）
skills/auto-tmux/scripts/auto-tmux.sh topology
skills/auto-tmux/scripts/auto-tmux.sh topology --session ai-hub枚举拓扑（原生命令兜底）
tmux list-sessions
tmux list-windows -a -F '#S:#I:#W#F'
tmux list-panes -a -F '#S:#I.#P #{pane_current_command} #{pane_title}'读取指定 pane 最近输出
skills/auto-tmux/scripts/auto-tmux.sh inspect -t <session>:<window>.<pane> -n 40
skills/auto-tmux/scripts/auto-tmux.sh capture -t <session>:<window>.<pane> -n 120原生读取指定 pane 最近 120 行
tmux capture-pane -t <session>:<window>.<pane> -p -S -120向指定 pane 发送命令或确认键
skills/auto-tmux/scripts/safety-check.sh --text "make test"
skills/auto-tmux/scripts/auto-tmux.sh send -t <session>:<window>.<pane> --text "make test" --enter
skills/auto-tmux/scripts/safety-check.sh --file /tmp/prompt.md --strict
skills/auto-tmux/scripts/auto-tmux.sh paste -t <session>:<window>.<pane> --file /tmp/prompt.md --enter --dry-run
skills/auto-tmux/scripts/auto-tmux.sh rescue -t <session>:<window>.<pane> --pattern "(y/n)" --reply y原生发送确认键
tmux send-keys -t <session>:<window>.<pane> "y" Enter安全中断挂住任务
tmux send-keys -t <session>:<window>.<pane> C-c开启/关闭窗口内广播（同步全部 pane）
tmux set-window-option synchronize-panes on
tmux set-window-option synchronize-panes off一次扫描全部窗口并截取最近输出（蜂群巡检）
skills/auto-tmux/scripts/auto-tmux.sh scan --session ai-hub -n 80
skills/auto-tmux/scripts/auto-tmux.sh scan --session ai-hub --pattern "ERROR|Traceback" --save-dir /tmp/auto-tmux-scan受控广播到指定 session
skills/auto-tmux/scripts/auto-tmux.sh broadcast --session ai-hub --text "pwd" --enter --dry-run生成证据快照/交接报告
skills/auto-tmux/scripts/auto-tmux.sh snapshot --session ai-hub --dir /tmp/auto-tmux-snapshot -n 120
skills/auto-tmux/scripts/swarm-brief.sh --session ai-hub --swarm-dir /tmp/ai_swarm --out /tmp/auto-tmux-brief
skills/auto-tmux/scripts/swarm-watch.sh --session ai-hub --swarm-dir /tmp/ai_swarm --iterations 3 --interval 10
skills/auto-tmux/scripts/swarm-archive.sh --session ai-hub --swarm-dir /tmp/ai_swarm --out /tmp/ai-hub-handoff.tar.gz
skills/auto-tmux/scripts/swarm-board.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-board.md
skills/auto-tmux/scripts/swarm-deps-graph.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-deps.md
skills/auto-tmux/scripts/swarm-export.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-export
skills/auto-tmux/scripts/swarm-timeline.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-timeline.md
skills/auto-tmux/scripts/swarm-blockers.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-blockers.md
skills/auto-tmux/scripts/swarm-results.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-results.md
skills/auto-tmux/scripts/swarm-results.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-results.md --jsonl /tmp/ai-swarm-results.jsonl
skills/auto-tmux/scripts/swarm-report-pack.sh --dir /tmp/ai_swarm --session ai-hub --out /tmp/ai-swarm-report-pack
skills/auto-tmux/scripts/swarm-report-pack.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-report-pack --attach /tmp/auto-tmux-remote
skills/auto-tmux/scripts/swarm-report-pack.sh --dir /tmp/ai_swarm --out /tmp/ai-swarm-report-pack --tar
skills/auto-tmux/scripts/swarm-assign.sh --swarm-dir /tmp/ai_swarm --session ai-hub --out /tmp/ai-swarm-assign.md
skills/auto-tmux/scripts/swarm-health.sh --session ai-hub --swarm-dir /tmp/ai_swarm --out /tmp/auto-tmux-health
skills/auto-tmux/scripts/remote-readonly.sh --host user@example.com --session ai-hub --out /tmp/auto-tmux-remote --dry-run远程救援：发现等待输入即发送 y
target="$(tmux list-panes -t ai-hub:worker1 -F '#S:#I.#P' | head -n 1)"
skills/auto-tmux/scripts/auto-tmux.sh rescue -t "$target" --pattern "(y/n)" --reply y --dry-run
skills/auto-tmux/scripts/auto-tmux.sh rescue -t "$target" --pattern "(y/n)" --reply y新建 AI 蜂群工作台（1 指挥 + 3 工人）
skills/auto-tmux/scripts/auto-tmux.sh hub --session ai-hub --workers 3 --cmd "codex"
skills/auto-tmux/scripts/auto-tmux.sh topology --session ai-hub
skills/auto-tmux/scripts/auto-tmux.sh cleanup --session ai-hub --dry-runCodex Pilot Mode（主 Codex 控制一个持久化交互式 worker Codex）
# 只在用户显式输入 Codex Pilot Mode 激活提示词后使用；完整提示词见 references/codex-pilot-mode.md。
tmux new-session -d -s codex-pilot -n worker 'codex'
bash skills/auto-tmux/scripts/auto-tmux.sh topology --session codex-pilot
bash skills/auto-tmux/scripts/auto-tmux.sh capture -t codex-pilot:worker.0 -n 120
bash skills/auto-tmux/scripts/swarm-dispatch.sh \
--role codex-worker \
--target codex-pilot:worker.0 \
--session codex-pilot \
--task "进入持久 worker 待命状态，等待主 Codex 后续分派任务" \
--send \
--dry-run启用 oh-my-tmux 配置（仓库内版本）
repo_root="$(git rev-parse --show-toplevel)"
ln -sfn "$repo_root/skills/auto-tmux/assets/oh-my-tmux/.tmux.conf" ~/.tmux.conf
cp -n "$repo_root/skills/auto-tmux/assets/oh-my-tmux/.tmux.conf.local" ~/.tmux.conf.local记录 pane 输出到文件
skills/auto-tmux/scripts/auto-tmux.sh record start -t <session>:<window>.<pane> --dir /tmp/auto-tmux-records
skills/auto-tmux/scripts/auto-tmux.sh record stop -t <session>:<window>.<pane>
skills/auto-tmux/scripts/record-summary.sh --dir /tmp/auto-tmux-records --out /tmp/auto-tmux-record-summary.md
skills/auto-tmux/scripts/check-jsonl.sh /tmp/ai-swarm-results.jsonl --require-key type --require-key id --require-key status
skills/auto-tmux/scripts/review-checklist.sh --pack /tmp/ai-swarm-report-pack --out /tmp/ai-swarm-report-pack/review-checklist.md
skills/auto-tmux/scripts/review-checklist.sh --pack /tmp/ai-swarm-report-pack --strict
skills/auto-tmux/scripts/verify-report-pack.sh --pack /tmp/ai-swarm-report-pack
skills/auto-tmux/scripts/incident-report.sh --type mis-send --out /tmp/auto-tmux-incident.md等待 pane 出现完成信号
skills/auto-tmux/scripts/auto-tmux.sh wait -t <session>:<window>.<pane> --pattern "Tests passed|BUILD SUCCESS" --timeout 300初始化蜂群状态、任务和锁
skills/auto-tmux/scripts/swarm-state.sh init --dir /tmp/ai_swarm
skills/auto-tmux/scripts/swarm-state.sh task-add --id task-001 --text "检查 README 链接"
skills/auto-tmux/scripts/swarm-state.sh task-import --file /tmp/tasks.txt --prefix batch --dir /tmp/ai_swarm
skills/auto-tmux/scripts/swarm-state.sh task-depend --id task-002 --blocked-by task-001 --dir /tmp/ai_swarm
skills/auto-tmux/scripts/swarm-state.sh task-ready --dir /tmp/ai_swarm
skills/auto-tmux/scripts/swarm-state.sh task-next --owner <session>:<window>.<pane>
skills/auto-tmux/scripts/swarm-state.sh lock-acquire --name README.md --owner <session>:<window>.<pane>
skills/auto-tmux/scripts/swarm-state.sh lock-prune --older-than 3600 --dry-run
skills/auto-tmux/scripts/swarm-state.sh task-block --id task-001 --owner <session>:<window>.<pane> --reason "等待输入"
skills/auto-tmux/scripts/swarm-state.sh task-fail --id task-001 --owner <session>:<window>.<pane> --reason "测试失败"
skills/auto-tmux/scripts/swarm-state.sh task-list --status FAIL --dir /tmp/ai_swarm
skills/auto-tmux/scripts/swarm-state.sh metrics --dir /tmp/ai_swarm
skills/auto-tmux/scripts/swarm-state.sh task-reopen --id task-001 --owner <session>:<window>.<pane> --reason "重新执行"
skills/auto-tmux/scripts/swarm-state.sh validate --dir /tmp/ai_swarm
skills/auto-tmux/scripts/swarm-state.sh report --dir /tmp/ai_swarm脚本端到端自测
source skills/auto-tmux/scripts/completion.bash
skills/auto-tmux/scripts/auto-tmux-smoke-test.sh
skills/auto-tmux/scripts/validate-auto-tmux.sh渲染 commander / worker / reviewer 提示词
skills/auto-tmux/scripts/render-swarm-prompt.sh commander --session ai-hub --task "拆分并推进任务"
skills/auto-tmux/scripts/render-swarm-prompt.sh worker --target <session>:<window>.<pane> --task "只处理指定子任务"
skills/auto-tmux/scripts/render-swarm-prompt.sh reviewer --session ai-hub --task "审查 worker 产出"
skills/auto-tmux/scripts/swarm-dispatch.sh --role worker --target <session>:<window>.<pane> --task "只处理指定子任务" --send --dry-runRules & Constraints
- MUST：优先使用
scripts/auto-tmux.sh；直接使用 tmux 原生命令时必须保留同等安全约束。 - MUST：在发送按键前用
capture-pane复核目标上下文；按键操作必须带<session>:<window>.<pane>绝对定位。 - MUST：遵循 oh-my-tmux 约定，不修改主配置文件；自定义写入
~/.tmux.conf.local。 - MUST：批量操作前先
list-windows/list-panes建立白名单，避免误控用户窗口。 - MUST：Codex Pilot Mode 只能在用户显式激活后进入；默认只控制一个持久化 worker Codex pane，且该 pane 必须对人类可见。
- MUST：Codex Pilot Mode 的 worker 必须是交互式
codex，禁止用无头codex exec替代。 - MUST：SSH/远程场景默认只读，先用
scripts/remote-readonly.sh采集证据；远程控制必须另行确认，不从只读脚本升级。 - SHOULD：救援/确认前先 grep 关键词（如
(y/n)、password），只对匹配目标发送。 - SHOULD：发送完整命令行时避免先发
Escape；先C-c中断、C-u清行，再用send-keys -l逐字发送完整命令。 - SHOULD：长 prompt、文件粘贴、跨 worker 分发前运行
scripts/safety-check.sh，批量发送先--dry-run。 - SHOULD：Codex Pilot Mode 中主 Codex 每次发送前先
inspect/captureworker pane，并把需求、上下文、边界、验收和验证命令整理成高质量提示词。 - SHOULD：pane 处在 Codex UI 时，先发送
/exit回到 shell 再执行命令。 - SHOULD：长任务开启
pipe-pane记录审计；广播完成后立即synchronize-panes off。 - SHOULD：多 worker 修改同一文件、目录或服务前，先用
swarm-state.sh lock-acquire声明锁。 - SHOULD：批量下发先用
broadcast --dry-run预览目标，再决定是否真实发送。 - NEVER：在未知 pane 发送破坏性命令；NEVER 在 root 会话不经确认发送
Ctrl+C/Ctrl+D。
Examples
Example 1：自动给卡住的安装脚本回车确认
- 输入：安装脚本停在
Proceed? (y/n)，先用topology找到真实 target。 - 步骤：
skills/auto-tmux/scripts/auto-tmux.sh topology找到目标<session>:<window>.<pane>。skills/auto-tmux/scripts/auto-tmux.sh rescue -t <target> --pattern "(y/n)" --reply y --dry-run确认命中。skills/auto-tmux/scripts/auto-tmux.sh rescue -t <target> --pattern "(y/n)" --reply y发送确认。
- 验收：脚本继续运行，无额外多余输入。
Example 2：蜂群巡检 + 自动救援
- 输入：4 个 AI 终端分布在 session
ai-hub的多个窗口。 - 步骤：
skills/auto-tmux/scripts/auto-tmux.sh scan --session ai-hub -n 80 --save-dir /tmp/auto-tmux-scan收集状态。- 对匹配
"(y/n)"的 pane 运行rescue --dry-run，确认后再真实发送。 - 对输出包含
Traceback或ERROR的 pane，先capture-pane记录再人工介入。
- 验收：等待确认的终端恢复，未误伤正常任务；巡检日志保存在本地。
Example 3：快速搭建 oh-my-tmux AI 工作台
- 输入：需要 1 个指挥 + 3 个 worker，并使用仓库内 oh-my-tmux 配置。
- 步骤：
- 运行「启用 oh-my-tmux 配置」片段完成软链和本地副本（若已存在跳过）。
skills/auto-tmux/scripts/auto-tmux.sh hub --session ai-hub --workers 3 --cmd "codex"创建 session。skills/auto-tmux/scripts/auto-tmux.sh topology --session ai-hub确认 worker 窗口存在。
- 验收：前缀键为
C-a，状态栏正常；4 个窗口可互相捕获/发送按键。
FAQ
- Q: 为什么按
C-b无反应？- A: oh-my-tmux 默认主前缀是
C-a，C-b仅作备用；确保无其他程序占用C-a。
- A: oh-my-tmux 默认主前缀是
- Q: 如何只广播当前窗口而不影响其他窗口？
- A: 只对当前窗口执行
set-window-option synchronize-panes on，使用完立即关闭。
- A: 只对当前窗口执行
- Q: 远程发送命令前如何确认当前工作目录？
- A:
tmux display-message -pt <target> '#{pane_current_path}'，必要时先send-keys 'cd /path' Enter。
- A:
Troubleshooting
no such session/window/pane→ 检查目标名称，先list-sessions/list-windows校验再重试。- 前缀键冲突 → 查看
~/.tmux.conf.local是否改写前缀；可在其中设置set -g prefix C-b #!important。 - 状态栏样式错乱/重复 → 按 README 建议用
tmux -f /dev/null -L test排查；多半是终端宽字符/字体导致。 - 发送命令没效果 → 目标 pane 可能在 copy-mode；优先用
tmux display-message -pt <target> '#{pane_in_mode}'判断，再用tmux send-keys -t <target> -X cancel退出；避免用Escape（会触发 Meta 前缀，可能吞掉首字符）。
Show full SKILL.md (311 more words)Show less
References
references/index.md: 导航与文件速览references/getting_started.md: 术语、oh-my-tmux 最小接入步骤references/api.md: tmux/oh-my-tmux 常用命令、选项与 gpakosz 特色键位references/automation.md:scripts/auto-tmux.sh子命令、安全模型与 AI 蜂群协作流程references/safety-policy.md: 发送、广播、清理、归档和敏感信息处理的安全策略references/session-safety.md: 本地 session、远程 SSH 和多 Agent 协作的分层安全边界references/jsonl-schema.md: JSONL/manifest 机器可读输出字段约定references/report-pack-review.md: report pack reviewer 最短验收路径references/reading-paths.md: commander、worker、reviewer、operator 的角色化阅读路线references/incident-runbook.md: 误发送、误广播、远程采集失败和敏感信息风险处理流程references/script-cheatsheet.md: 观察、控制、协作、报告、审计和事故处理脚本速查references/swarm-state.md: 蜂群状态、任务、锁和报告协议references/prompt-templates.md: commander/worker/reviewer 提示词模板和下发方式references/codex-pilot-mode.md: 主 Codex 控制一个持久化交互式 worker Codex 的激活提示词、边界和验收协议references/ai-swarm-collaboration.md: tmux 蜂群协作历史文档、架构模式、协议、案例和风险限制references/iteration-roadmap.md: 多轮迭代记录、能力层和后续候选方向references/iteration-closeout.md: 50 轮迭代后的能力收尾、验证链路和维护边界references/examples.md: 蜂群协议脚本与长示例references/troubleshooting.md: 典型故障到修复路径scripts/auto-tmux.sh: 安全封装的 tmux 自动化脚本入口scripts/swarm-state.sh: 蜂群任务、锁、状态和报告脚本scripts/swarm-brief.sh: 只读生成 tmux 蜂群交接报告scripts/swarm-watch.sh: 有限轮次巡检 tmux 蜂群输出和状态报告scripts/swarm-archive.sh: 打包 brief、snapshot 和 swarm state 的交接归档scripts/swarm-board.sh: 渲染任务、依赖、锁和状态日志的 Markdown 看板scripts/swarm-deps-graph.sh: 渲染deps.tsv的 Mermaid 依赖图scripts/swarm-export.sh: 导出任务、依赖和锁的 JSONL 状态包scripts/swarm-timeline.sh: 渲染status.log的 Markdown 时间线scripts/swarm-blockers.sh: 聚合 BLOCKED/FAIL 任务、锁和最近异常状态scripts/swarm-results.sh: 汇总 DONE/FAIL/BLOCKED worker 结果，支持 JSONL 导出scripts/swarm-report-pack.sh: 聚合 board、依赖图、时间线、阻塞清单、分配建议、附件、manifest.json和可选 tar.gzscripts/swarm-assign.sh: 根据 ready tasks 和 worker pane 生成只读分配建议scripts/swarm-health.sh: 汇总 tmux 与 swarm state 的只读健康检查报告包scripts/remote-readonly.sh: 通过 SSH 只读采集远端 tmux 拓扑、pane 输出和metadata.jsonlscripts/record-summary.sh: 汇总 pane 录制日志并生成复盘摘要scripts/check-jsonl.sh: 轻量检查 JSONL 行和必需字段scripts/review-checklist.sh: 为 report pack 生成 reviewer 审计清单scripts/verify-report-pack.sh: 统一验证 report pack 的 manifest、JSONL 和审计清单scripts/incident-report.sh: 生成误发送、误广播、远程采集和敏感信息风险复盘模板scripts/audit-package.sh: 检查脚本、参考资料、SKILL 入口和 validator 的索引一致性scripts/completion.bash: Bash completion，补全auto-tmux.sh与swarm-state.shscripts/safety-check.sh: 发送/粘贴/分发前检查危险命令、敏感信息和过大 payloadscripts/render-swarm-prompt.sh: commander/worker/reviewer/codex-worker 提示词渲染脚本scripts/swarm-dispatch.sh: 渲染并可选下发 commander/worker/reviewer 提示词scripts/auto-tmux-smoke-test.sh: tmux 自动化脚本端到端自测scripts/validate-auto-tmux.sh: auto-tmux 专属质量门禁assets/oh-my-tmux: gpakosz/oh-my-tmux submodule 的相对软链接入口assets/tmux-src: tmux/tmux submodule 的相对软链接入口
Maintenance
- Sources:
skills/auto-tmux/assets/oh-my-tmux、skills/auto-tmux/assets/tmux-src、蜂群实战提示词、Skill Seeker 生成的抓取材料（如需扩展自动化）。 - Last updated: 2026-05-19
- Known limits: 未涵盖 tmux 插件管理（tpm）及多用户共享权限配置；需时再补充。
Quality Gate
发布前自检（配合 auto-skill 校验脚本）：
description已含“capture-pane/send-keys/蜂群”等触发关键词。- 「When to Use」「Not For」已覆盖边界；前缀/版本要求清晰。
- Quick Reference ≤ 20 条且均可直接执行；无长篇解释。
- ≥3 个端到端示例，含输入/步骤/验收。
- 长文档放在
references/并可导航；无文档堆砌。 - 不确定项给出验证路径；禁止虚构 tmux 行为。
bash -n skills/auto-tmux/scripts/auto-tmux.sh、bash -n skills/auto-tmux/scripts/swarm-state.sh、bash -n skills/auto-tmux/scripts/swarm-brief.sh、bash -n skills/auto-tmux/scripts/swarm-watch.sh、bash -n skills/auto-tmux/scripts/swarm-archive.sh、bash -n skills/auto-tmux/scripts/swarm-board.sh、bash -n skills/auto-tmux/scripts/swarm-deps-graph.sh、bash -n skills/auto-tmux/scripts/swarm-export.sh、bash -n skills/auto-tmux/scripts/swarm-timeline.sh、bash -n skills/auto-tmux/scripts/swarm-blockers.sh、bash -n skills/auto-tmux/scripts/swarm-results.sh、bash -n skills/auto-tmux/scripts/swarm-report-pack.sh、bash -n skills/auto-tmux/scripts/swarm-assign.sh、bash -n skills/auto-tmux/scripts/swarm-health.sh、bash -n skills/auto-tmux/scripts/remote-readonly.sh、bash -n skills/auto-tmux/scripts/record-summary.sh、bash -n skills/auto-tmux/scripts/check-jsonl.sh、bash -n skills/auto-tmux/scripts/review-checklist.sh、bash -n skills/auto-tmux/scripts/verify-report-pack.sh、bash -n skills/auto-tmux/scripts/completion.bash、bash -n skills/auto-tmux/scripts/safety-check.sh、bash -n skills/auto-tmux/scripts/render-swarm-prompt.sh、bash -n skills/auto-tmux/scripts/swarm-dispatch.sh与bash -n skills/auto-tmux/scripts/validate-auto-tmux.sh通过。skills/auto-tmux/scripts/auto-tmux-smoke-test.sh通过或在无 tmux 环境下明确跳过。- 运行
skills/auto-tmux/scripts/validate-auto-tmux.sh和skills/auto-skill/scripts/validate-skill.sh skills/auto-tmux --strict通过。
© tradecatlabs, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 57 other files (scripts, references, assets) in skills/auto-tmux of tradecatlabs/vibe-coding-cn.
- SKILL.md
- AGENTS.md
- README.md
- assets/AGENTS.md
- assets/README.md
- assets/oh-my-tmux
- assets/tmux-src
- references/AGENTS.md
- references/README.md
- references/ai-swarm-collaboration.md
- references/api.md
- references/automation.md
- references/codex-pilot-mode.md
- references/examples.md
- references/getting_started.md
- references/incident-runbook.md
- references/index.md
- references/iteration-closeout.md
- references/iteration-roadmap.md
- … and 39 more
Open the folder on GitHubat commit 5b76a8f
Compare with similar skills
Auto tmux Operator next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Auto tmux Operator this skilltradecatlabs/vibe-coding-cn | 17k | — | ~4.7k | Automated safety check: Pass | MIT | |
| OpenRig Upgrade Proceduremvschwarz/openrig | 6.6k | — | ~2.9k | Automated safety check: Pass | Apache-2.0 | |
| Pilot AWS Daemon Operationsqf-studio/pilot | 722 | — | ~3k | Automated safety check: Notes | Custom licence | |
| Doca Bare Metal DeploymentNVIDIA/skills | 3.6k | — | ~2.9k | Automated safety check: Pass | Apache-2.0 | |
| Agent of Empires Session Manageragent-of-empires/agent-of-empires | 3.3k | — | ~2.5k | Automated safety check: Pass | MIT | |
| Oh My OpencodeLeoYeAI/openclaw-master-skills | 2.2k | — | ~5.1k | Automated safety check: Notes | MIT |
Similar skills
Pilot AWS Daemon Operations
qf-studio/pilot
Checks status, tails logs, queries the task queue, and starts, stops or restarts the Pilot daemon running on a specific AWS EC2 instance over SSM.
- Official
Doca Bare Metal Deployment
NVIDIA/skills
A skill your agent uses for launching, supervising, debugging, OR platform lifecycle on a BlueField — BFB install, RShim/TMFIFO, host PF rebind, post-BFB recovery — taking a DOCA-linked binary to a…
Agent of Empires Session Manager
agent-of-empires/agent-of-empires
Launches, monitors and organizes AI coding agent sessions such as Claude Code or Codex inside tmux, tracking status, capturing output and managing git worktrees for parallel branches.
Oh My Opencode
LeoYeAI/openclaw-master-skills
Multi-agent orchestration plugin for OpenCode. An agent skill from LeoYeAI/openclaw-master-skills.
Agent of Empires Session Manager
agent-of-empires/agent-of-empires
Starts, monitors and organizes coding agent sessions that run in tmux through the aoe command, including groups, profiles and worktree-based parallel work.
More from tradecatlabs/vibe-coding-cn
All 17 skills in this repoAuto Skill Builder
tradecatlabs/vibe-coding-cn
Meta-skill that turns docs, APIs, code or specs into a reusable skill with references and a quality gate, and refactors skills that are unclear or misfire.
Web3 Smart Contract Grep Arsenal
tradecatlabs/vibe-coding-cn
A master set of ten grep command blocks that surface likely vulnerability classes in Solidity source within the first 30 minutes of auditing a new protocol.
Web3 Bug Bounty AI Tools
tradecatlabs/vibe-coding-cn
A selection guide to AI-driven tools for Web3 bug bounty work, from autonomous web pentesters to smart contract bug finders, with notes on authorization.
Ethereum Smart Contract Vulnerability Analysis
tradecatlabs/vibe-coding-cn
Runs Slither and Mythril against Solidity contracts to find reentrancy, overflow and access-control bugs before mainnet deployment, then triages and reports findings.
Math Computation
tradecatlabs/vibe-coding-cn
Runs reproducible math computations and counterexample searches with SymPy, NumPy and mpmath, logging evidence without presenting results as proofs.
DeFi Smart Contract Bug Classes
tradecatlabs/vibe-coding-cn
Reference for ten classes of DeFi smart contract bugs, each with root cause, vulnerable code, fix, grep patterns and paid examples, for audits and bug bounty reviews.
Related
Works with
Categories
Questions about Auto tmux Operator
What does Auto tmux Operator do?
Operates tmux sessions like an administrator: reads pane output, sends keys, inspects many panes at once, and coordinates multiple AI terminals through a swarm state script, built on oh-my-tmux. sh wrapper over raw tmux commands for reading a pane's recent output, sending keystrokes, batch-inspecting several panes, and rescuing or recording a stuck terminal, falling back to native tmux commands such as capture-pane and send-keys only when needed.sh script manages a task queue, locks, status and results for coordinating multiple AI terminals, and the skill ships a vendored copy of the oh-my-tmux configuration under assets so its double-prefix shortcuts and pane sync commands are available without touching the user's own tmux config.
When should I use Auto tmux Operator?
Auto tmux Operator fits situations like: reading recent output from one or many tmux panes at once; sending confirmation keystrokes or commands to a specific pane safely; coordinating several AI agent terminals through a shared task queue; rescuing a stuck tmux pane without guessing at its current state.
How do I install Auto tmux Operator in Claude Code?
Run `npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a claude-code`. Or copy the skill folder (skills/auto-tmux in tradecatlabs/vibe-coding-cn) into .claude/skills/auto-tmux in your project. Claude Code loads it when a task matches its description.
How do I install Auto tmux Operator in Codex?
Run `npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a codex`. Or copy the skill folder (skills/auto-tmux in tradecatlabs/vibe-coding-cn) into .agents/skills/auto-tmux in your project. Codex loads it when a task matches its description.
Can I use Auto tmux Operator in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add tradecatlabs/vibe-coding-cn --skill auto-tmux -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/auto-tmux, .gemini/skills/auto-tmux, .github/skills/auto-tmux and .opencode/skills/auto-tmux in your project.
What does Auto tmux Operator need to run?
Going by SKILL.md and its folder, Auto tmux Operator needs the command-line tools its instructions call (bash, git and codex). Our summary lists: tmux 2.6 or later; Bash.
Does Auto tmux Operator access the network?
SKILL.md contains no URLs. Its commands use git, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is Auto tmux Operator safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
What licence does Auto tmux Operator use?
Auto tmux Operator is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Auto tmux Operator use?
About 4.7k tokens (SKILL.md is roughly 19k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 20k tokens, read only when the agent opens those files.
What are the alternatives to Auto tmux Operator?
Skills that share tags, products or a category with Auto tmux Operator: OpenRig Upgrade Procedure (mvschwarz/openrig, 6.6k stars), Pilot AWS Daemon Operations (qf-studio/pilot, 722 stars), Doca Bare Metal Deployment (NVIDIA/skills, 3.6k stars) and Agent of Empires Session Manager (agent-of-empires/agent-of-empires, 3.3k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Auto tmux Operator?
tradecatlabs (a GitHub user) maintains it in tradecatlabs/vibe-coding-cn, which has 17,386 GitHub stars. The repository holds 17 skills in this directory. The repository was last updated on October 10, 2026.
Source: tradecatlabs/vibe-coding-cn on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.