Claude Code Orchestrator (tmux-first) for Openclaw

A robust orchestration framework for running Claude Code in observable tmux sessions with automated progress tracking and completion callbacks.

yaxuan42
v0.2.0
Feb 15, 2026
0
1.6k
58

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install claude-code-orchestrator

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install claude-code-orchestrator 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 Claude Code Orchestrator (tmux-first)?

This skill provides a specialized framework for executing long-running AI coding tasks via Claude Code within stable tmux sessions. By leveraging tmux-based orchestration, developers can prevent silent hangs common in standard shell executions and maintain full visibility into the agent's progress. It is designed for complex development workflows where an agent needs to work asynchronously while the developer remains informed.

Integration with Openclaw Skills allows for a hands-off approach where the orchestrator manages the session lifecycle. It handles everything from prompt submission to automated completion reports, ensuring that the developer receives a comprehensive analysis of changes, risks, and next steps once the task is finished.

Claude Code Orchestrator (tmux-first) Use Cases

  • Running complex, multi-file refactoring tasks that require significant processing time.
  • Monitoring AI agent progress over SSH or remote environments without risk of session loss.
  • Automating background coding jobs that notify the developer upon completion via a wake signal.
  • Generating structured status summaries for multiple concurrent AI coding tasks.

How Claude Code Orchestrator (tmux-first) Works

  1. The orchestrator creates a dedicated prompt file to handle large contexts and avoid shell quoting limitations.
  2. A managed tmux session is initiated using a specific socket to host the interactive Claude environment.
  3. The skill launches Claude with the dangerously-skip-permissions flag and programmatically pastes the task prompt.
  4. Progress is monitored via bash scripts that can capture pane output or check the session status (running, stuck, idle, or dead).
  5. Upon completion, the task executes a wake script that sends a callback to Openclaw Skills.
  6. A mandatory completion loop reads the generated report and tmux transcript to provide the user with a technical risk assessment and summary.

Claude Code Orchestrator (tmux-first) Setup

Ensure your environment has tmux installed and the orchestration scripts are placed in your base directory. Configure your workspace paths accordingly.

# Start a new coding task with a specific label
bash {baseDir}/scripts/start-tmux-task.sh \
  --label "feature-update" \
  --workdir "/your/project/path" \
  --prompt-file "/path/to/instructions.md"
# Monitor an active session
bash {baseDir}/scripts/monitor-tmux-task.sh --attach --session cc-feature-update

Claude Code Orchestrator (tmux-first) Data Schema & Taxonomy

The skill manages task metadata and session state through a structured filesystem and tmux socket taxonomy.

Component Details
Session Labels Prefixed with cc- for easy identification within tmux.
Completion Reports Stored as JSON/MD at /tmp/cc-<label>-completion-report.json.
Tmux Socket Located by default at $TMPDIR/clawdbot-tmux-sockets/clawdbot.sock.
Task Status Categorized as running, likely_done, stuck, idle, dead, or done_session_ended.

Claude Code Orchestrator (tmux-first) Advanced Features

  • Zero-token status checks to determine task health without consuming LLM resources.
  • Remote host support for listing and managing sessions on remote servers via SSH aliases.
  • Robust submission logic with ready-checks and multi-Enter retries to ensure prompts are processed.
  • Butler-style summaries that pipe JSON task lists into Openclaw Skills for periodic status updates.
  • Automated recovery commands to reconstruct evidence and generate reports if a session is lost.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*