Coding Sessions for Openclaw

Orchestrate persistent AI coding agents in tmux sessions with automatic notifications and health monitoring.

luke-deltadesk
v1.0.0
Mar 4, 2026
0
1.2k
0

Install & Download

1. ClawHub CLI

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

npx clawhub@latest install coding-sessions

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 coding-sessions 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 Coding Sessions?

Coding Sessions provides a robust framework for running AI-driven development agents like Codex and Ralph in the background. By leveraging persistent tmux sessions via a stable socket, this skill ensures that intensive coding tasks survive process restarts or gateway interruptions.

This system integrates seamlessly with Openclaw Skills to provide real-time event notifications upon task completion, allowing developers to manage multiple parallel workstreams without manual supervision. It is designed specifically for tasks exceeding five minutes, where standard process execution is prone to failure.

Coding Sessions Use Cases

  • Running multi-step coding tasks that require extended execution times.
  • Executing Ralph loops with PRD validation for complex feature development.
  • Orchestrating parallel independent coding agents to speed up project delivery.
  • Managing background programming sessions that must survive system restarts or shell exits.

How Coding Sessions Works

  1. The system initializes a tmux session using a stable socket (~/.tmux/sock) to prevent session reaping by the OS.
  2. It configures the environment PATH to ensure the AI agents have access to necessary Homebrew and CLI binaries.
  3. The designated AI agent (Codex or Ralph) is launched within the isolated session with project-specific context.
  4. A completion hook captures the exit code as soon as the agent finishes its task.
  5. The skill triggers an Openclaw system event to provide an instant status notification.
  6. The session enters a long sleep state, preserving the terminal output for manual verification and logging.

Coding Sessions Setup

Ensure tmux is installed on your system. To launch a persistent session for a Codex task, use the following command structure:

tmux -S ~/.tmux/sock new -d -s <name> "cd <project-dir> && PATH=/opt/homebrew/bin:\$PATH codex exec --full-auto '<task-description>'; EXIT_CODE=\$?; openclaw system event --text '<name> finished (exit \$EXIT_CODE)' --mode now; sleep 999999"

For complex work, use the Ralph loop configuration:

tmux -S ~/.tmux/sock new -d -s <name> "cd <project-dir> && PATH=/opt/homebrew/bin:\$PATH ralphy --codex --prd PRD.md; EXIT_CODE=\$?; openclaw system event --text 'Ralph loop <name> finished' --mode now; sleep 999999"

Coding Sessions Data Schema & Taxonomy

The skill organizes execution data and progress tracking through the following structures:

Data Component Description
PRD Checklists Markdown files with GFM checkboxes used by Ralph to track task completion.
Stable Socket Located at ~/.tmux/sock to ensure process persistence.
Git History The primary source of truth for verifying code changes made by the agents.
Daily Notes External tracking in memory/YYYY-MM-DD.md to monitor active long-running processes.

Coding Sessions Advanced Features

  • Support for parallel Ralph loops to handle independent architectural tasks simultaneously.
  • Automated completion hooks that integrate directly with the Openclaw Skills event system.
  • Use of stable sockets to bypass macOS /tmp directory cleanup cycles.
  • Post-completion verification protocols involving git diff analysis and log tailing.
  • Automatic retry logic when wrapping agents in Ralph loops to handle context stalls.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*