launch

Launch autonomous workspace sessions with jj and Claude Code in tmux.

14|Updated Mar 23, 2026

Search Tags:#agent #workspace #tmux #claude-code #launch #orchestrate #cmux

Ready to run

npx skills add https://github.com/memorysaver/agentic-engineering-patterns --skill launch-memorysaver

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: launch
Source: https://github.com/memorysaver/agentic-engineering-patterns/tree/main/skills/agentic-development-workflow/launch
Command: npx skills add https://github.com/memorysaver/agentic-engineering-patterns --skill launch-memorysaver

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Spawn an autonomous workspace session to implement a feature. This step helps ensure a clean, reproducible environment by bootstrapping a jj workspace and a Claude Code agent in a tmux/cmux session, with an optional evaluator agent for quality assurance.

Core Features & Use Cases

Creates a JJ workspace outside sensitive areas to avoid permissions prompts and keep work isolated.
Boots a Claude Code agent in a tmux/cmux session and optionally starts a separate evaluator agent for quality assurance.
Fits into the /design → /launch → /build workflow to accelerate feature delivery with parallel execution.

Quick Start

Create a new autonomous workspace and boot Claude Code in a tmux session to begin implementing the selected feature.

Frequently Asked Questions about launch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I spawn an autonomous workspace to implement a feature with Claude Code?▼

To spawn an autonomous workspace, you bootstrap a jj workspace and launch a Claude Code agent in a tmux session. This creates a reproducible environment for parallel feature development.

Does spawning a Claude Code agent require a clean working copy?▼

Yes, spawning a Claude Code agent requires a clean working copy and remote dispatch readiness. A configured calibration context is also needed to ensure reproducible and safe agent execution.

Can I run an evaluator agent for quality assurance in a tmux session?▼

Yes, you can optionally start a separate evaluator agent alongside the Claude Code agent in a tmux session. This provides quality assurance during the feature implementation process.

How do I isolate feature development to avoid permissions prompts?▼

You isolate feature development by creating a jj workspace outside sensitive areas. This approach keeps work isolated and prevents permissions prompts during autonomous agent execution.

When should I not use autonomous workspace sessions for feature development?▼

You should not use autonomous workspace sessions if you lack a configured calibration context or a clean working copy. These prerequisites are required for reproducible and safe agent execution.

What is the best way to orchestrate parallel feature implementation with Claude Code?▼

The best way to orchestrate parallel feature implementation is using the /design to /launch to /build workflow. Launch bootstraps a jj workspace and a Claude Code agent in a cmux environment to accelerate delivery.