A powerful terminal multiplexing skill that enables Openclaw agents to remote-control interactive CLI sessions and scrape TTY output.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install tmux-temp
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install tmux-temp using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is tmux?
The tmux skill provides a robust interface for Openclaw Skills to interact with terminal-based applications that require a persistent or interactive TTY. Unlike standard background execution, this skill allows for granular control over terminal panes, enabling the injection of keystrokes and the capturing of screen buffers. This makes it an essential tool for managing complex CLI environments like REPLs, interactive debuggers, or long-running processes that need to be monitored in real-time.
By leveraging isolated sockets, this skill ensures that automated workflows do not interfere with a user's personal tmux sessions. It transforms the terminal into a programmable environment where AI agents can execute commands, wait for specific textual prompts, and retrieve historical output for analysis, providing a bridge between static automation and dynamic command-line interaction.
tmux Use Cases
- Running and controlling interactive Python or Node.js REPLs programmatically.
- Orchestrating multiple coding agents like Claude Code or Codex to perform parallel tasks across different git worktrees.
- Monitoring long-running build or test processes by scraping pane history for specific success or failure patterns.
- Debugging automated workflows by manually attaching to isolated tmux sessions to inspect the live state.
- Automating CLI tools that do not support non-interactive modes by simulating human keystrokes.
How tmux Works
- The skill initializes an isolated tmux socket within a dedicated directory to ensure a clean execution environment.
- A new detached session or window is created, targeting a specific task or shell environment.
- Commands are sent to the terminal using the send-keys functionality, supporting both literal strings and control sequences like C-c.
- The agent monitors the output by capturing the pane's buffer or using specialized scripts to poll for specific regex patterns.
- Upon detecting a completion prompt or required text, the agent extracts the relevant data from the terminal history.
- Finally, the session or the entire socket server is terminated to free up system resources.
tmux Setup
The tmux skill requires tmux to be installed on your system (macOS or Linux/WSL). Configure the socket directory to manage isolated sessions for Openclaw Skills.
# Define the socket directory
export CLAWDBOT_TMUX_SOCKET_DIR="${TMPDIR:-/tmp}/clawdbot-tmux-sockets"
mkdir -p "$CLAWDBOT_TMUX_SOCKET_DIR"
# Initialize a session on a private socket
SOCKET="$CLAWDBOT_TMUX_SOCKET_DIR/clawdbot.sock"
tmux -S "$SOCKET" new -d -s openclaw-session
tmux Data Schema & Taxonomy
The skill organizes terminal data through a hierarchical structure of sockets, sessions, and panes.
| Component | Description | Example Format |
|---|---|---|
| Socket Path | The file system path to the tmux communication socket | /tmp/clawdbot.sock |
| Target ID | The specific identifier for a session, window, and pane | session:window.pane (e.g., agent-1:0.0) |
| Pane Buffer | The captured text output from the terminal screen | UTF-8 String (Plain Text) |
| Metadata | Information about active sessions and pane dimensions | JSON / Key-Value Pairs |
tmux Advanced Features
- Isolated Socket Management: Run hundreds of parallel sessions without polluting the default tmux server.
- wait-for-text Helper: A sophisticated polling script that allows agents to wait for specific terminal states with configurable timeouts and intervals.
- Parallel Agent Orchestration: Launch and monitor dozens of AI coding agents simultaneously across different workdirs using a single socket.
- Python REPL Optimization: Automatic environment configuration (PYTHON_BASIC_REPL=1) to ensure compatibility with keystroke injection.
- Pane Scraping: Advanced history capture capable of retrieving thousands of lines of terminal output for deep context analysis.
SKILL.md
Loading
Related Openclaw Skills
Read It Later
An intelligent content management skill that saves and organizes links, articles, and notes into a structured markdown-based reading list.
Artifacts Builder
A professional suite of tools designed to build, bundle, and share complex React-based Claude.ai artifacts with shadcn/ui integration.
Local File Reader
A powerful utility for AI agents to access, summarize, and search content across various local file formats.
Restaurant Review Cross-Check
A data-driven validation tool that cross-references restaurant ratings and social engagement across Xiaohongshu and Dianping to ensure recommendation accuracy.
Empirical Paper Analysis Skill
A specialized AI tool for deep, structured analysis of empirical research papers focusing on causal inference and methodological rigor.
Financial Times Deep Reader (ft-reader)
Automate the extraction, synthesis, and bilingual reporting of Financial Times articles for professional and academic research.