Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Security + audit fixes: declare requirements, remove forced proxy, lint/build opt-in
What This Skill Does
Orchestrates Claude Code development tasks inside tmux sessions, keeping progress visible and ensuring a completion callback fires when the task finishes. Long-running coding work runs in a named session that can be attached, monitored, or inspected remotely. On completion, a structured report is generated and analyzed before delivering results.
Unlike a one-shot claude -p call that can hang silently, tmux sessions keep output visible and the mandatory wake callback ensures analysis happens after confirmed completion rather than polling.
When to use it
- Running an overnight refactor and getting notified when Claude Code finishes
- SSHing into a remote machine to check on a running coding task
- Listing all active Claude Code sessions and their current status at a glance
- Recovering a stuck or dead session without losing task context
- Reviewing a structured diff and risk summary after a large automated code change
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Run a gallery website redesign task using Claude Code and notify me when done
- 1Write task instructions to a prompt file to avoid shell quoting issues
- 2Start a named tmux session via start-tmux-task.sh with the target workdir and prompt file
- 3Launch Claude Code in interactive mode and paste the prompt with a wake callback command embedded
- 4Return the tmux attach command so the user can observe progress over SSH
- 5On wake event, read the completion report JSON and tmux transcript, then provide diff summary and risk analysis
Session name, attach command, and a structured completion analysis covering what changed, reliability of the output, and concrete next steps
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.