watch-parallel-stacks-cursor

To coordinate parallel worktrees for stacked PRs, you need a durable watcher and per-lane workers that process a JSON plan with issue ids, keeping dependent tickets in the same worktree. This ensures dependency order is preserved while independent lanes progress concurrently.

Dependency-aware parallel orchestration with tmux works by generating a long-lived watcher session and separate worker sessions for each active lane. The watcher validates plan dependencies and maintains a state.json file, relaunching workers as needed until all planned tasks are complete.

Yes, you need tmux installed to run parallel lane workers for multi-ticket implementation. The tool generates tmux-backed watcher and lane sessions to manage the concurrent execution and coordinate the dependency graph across independent workstreams.

The best way to manage stacked PR sequences for dependent Linear tickets is to group directly dependent tickets in the same lane and worktree. This allows dependent changes to land as an intentional stacked PR sequence while independent tickets proceed in parallel.

To create a JSON plan for dependency-aware parallel task execution, define your Linear issue ids, assign them to specific lanes, and map their dependencies. The launcher command reads this plan.json file to validate dependencies and generate the tmux sessions.

You should not use parallel lane orchestration for PR stacking when your tickets lack clear dependency graphs or cannot be grouped into independent lanes. It requires that independent work can proceed concurrently, while blocked and dependent tickets must wait or share a worktree.