parallelize-work-cursor

Orchestrate parallel Cursor agent work via wt worktrees and tmux sessions.

Updated May 18, 2025
One-click install
npx skills add https://github.com/nthpaul/dotfiles --skill parallelize-work-cursor
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: parallelize-work-cursor
Source: https://github.com/nthpaul/dotfiles/tree/main/cursor/.cursor/skills/parallelize-work-cursor
Command: npx skills add https://github.com/nthpaul/dotfiles --skill parallelize-work-cursor

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires wt, tmux, zsh, agent, and includes scripts (resource) components.

What problem does it solve?

Coordinating multiple independent implementation tasks serially wastes time and delays merges when your work can be safely split across concurrent Linear issues.

Core Features & Use Cases

  • Per-issue worktree isolation: creates one wt worktree per explicitly identified Linear issue, preventing cross-talk between tasks.
  • Detached Cursor agent sessions: starts a detached Cursor agent session per task inside the matching worktree so you can attach and steer each run independently.
  • Explicit parallel-safety from source-of-truth: only parallelizes tasks that are clearly marked safe in an input document or spec, avoiding guesswork about dependencies.

Use case: You have an implementation plan or TRD that names multiple independent Linear issues (e.g., frontend work and backend work that do not block each other); run this skill to spawn agents for each issue concurrently and keep their scopes constrained.

Quick Start

Run: call launch_parallel_ticket_work.py with a plan JSON that includes a required source doc URL/path and a tasks array mapping each parallel Linear issue to its project and worktree name.

Frequently Asked Questions about parallelize-work-cursor

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

FAQPage Schema
How do I run multiple Cursor agents in parallel for separate Linear issues?

You can parallelize Cursor agents by providing a JSON plan mapping each explicitly independent Linear issue to a specific project and worktree, which the script uses to spawn isolated, concurrent agent sessions.

What do I need to set up before launching parallel agent workflows with tmux and wt?

You must provide a JSON implementation plan with a required source document, optional spec, and validated mappings for issue_id, project, and worktree, alongside having wt, tmux, zsh, and agent installed.

Can I use this to parallelize any implementation tasks in my TRD?

No, it only parallelizes tasks explicitly marked as independent in your source-of-truth document, avoiding guesswork about dependencies and ensuring each task maps to exactly one Linear issue and one wt project.

How does wt worktree isolation prevent cross-talk between concurrent Cursor agents?

It creates one dedicated wt worktree per Linear issue, keeping each concurrent Cursor agent session strictly scoped to its own directory and preventing file or state cross-talk between parallel task executions.

How do I monitor and steer concurrent Cursor agent sessions running in tmux?

The skill starts detached tmux sessions per task with generated prompts and log capture, allowing you to independently attach to any session to monitor output and steer the Cursor agent running in that specific worktree.