diverge

Generate multiple implementation directions with executable launcher scripts.

Updated Dec 19, 2025
One-click install
npx skills add https://github.com/Qnurye/dotfiles --skill diverge-qnurye
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: diverge
Source: https://github.com/Qnurye/dotfiles/tree/main/agents/skills/diverge
Command: npx skills add https://github.com/Qnurye/dotfiles --skill diverge-qnurye

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires jq, wt, claude, git, shasum, and includes scripts (resource) components.

What problem does it solve?

Diverge enables teams to explore multiple implementation directions for a goal by grounding research and iteratively refining options with user input, then producing executable launcher scripts for each chosen approach.

Core Features & Use Cases

  • Grounded goal exploration: generate several viable directions based on the goal and the available context.
  • Parallel launcher generation: create ready-to-run scripts to bootstrap implementation in isolated workspaces.
  • Interactive refinement: incorporate user feedback to prune or combine directions before execution.

Quick Start

Describe your goal and use the Diverge skill to generate launcher scripts for multiple directions.

Frequently Asked Questions about diverge

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

FAQPage Schema
How do I explore multiple software engineering implementation directions for a single goal?

Divergent planning with parallel launcher scripts generates several viable implementation directions based on your goal and context. It produces ready-to-run scripts that bootstrap each approach in isolated worktrees for parallel exploration.

How do I generate launcher scripts to bootstrap parallel implementation in isolated worktrees?

You describe your goal and grounded context, then refine the proposed directions interactively. The tool generates ready-to-run launcher scripts for each chosen approach, which can be executed directly in isolated git worktrees.

What's the best way to refine multiple implementation directions before committing to one?

Interactive refinement allows you to incorporate user feedback to prune or combine directions before execution. This ensures you only generate launchers for approaches that have been iteratively validated against your grounded input context.

Do I need git and worktree installed to run parallel launcher scripts for software planning?

Yes, you need git, wt, jq, shasum, and claude installed. These dependencies are required to generate formal plans per direction and execute the ready-to-run launcher scripts in isolated workspaces.

Can I use this parallel planning approach for non-software-engineering automation tasks?

This tool applies specifically to software engineering planning scenarios where parallel exploration is valuable. It requires grounded input context and produces formal plans with executable scripts tailored for isolated worktree environments.

Why does parallel planning require a grounded input context before generating directions?

Grounded input context ensures the proposed implementation directions are viable and tailored to your specific goal. Without it, the formal plans and generated launcher scripts cannot accurately bootstrap implementation in isolated workspaces.