mad-parallel

What problem does it solve?

Manually coordinating parallel feature development across multiple git worktrees requires tedious tracking of dependencies, file ownership, and merge order, often leading to merge conflicts, wasted rework, and delayed milestone delivery. This Skill automates the entire end-to-end workflow for wave-based parallel feature implementation, eliminating manual coordination overhead and reducing integration risks for teams using the MAD development framework.

Core Features & Use Cases

• Automated Dependency & Wave Orchestration: Analyzes feature dependencies to build a directed acyclic graph (DAG), assigns features to sequential execution waves to avoid conflicts, and automatically splits waves into sub-waves when they exceed the maximum allowed parallel feature limit.
• Isolated Worktree Setup: Creates independent git worktrees for each feature with properly linked shared project resources, so implementers can work in parallel without interfering with each other or the stability of the main branch.
• Integrated MAD Pipeline Execution: Runs the full planning, implementation, and validation workflow per feature, with native agent team support for parallel code implementation and automatic sequential fallback for environments without agent team access.
• Use Case: A team implementing 6 independent features for a new milestone can use this Skill to automatically split them into 2 waves of 3 features each, set up isolated worktrees for each feature, run planning for all features, dispatch parallel implementers for the first wave, validate and merge each feature, then repeat the process for the second wave, cutting total implementation time by 3x compared to sequential development.

Quick Start

Provide the path to the milestone map generated by the mad-decompose skill to automatically orchestrate parallel implementation of all mapped features across isolated git worktrees.