dwr

Orchestrate end-to-end development cycles across isolated subagents with wave-based parallelism.

Updated Mar 21, 2023
One-click install
npx skills add https://github.com/bagrinsergiu/MB-migration --skill dwr
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: dwr
Source: https://github.com/bagrinsergiu/MB-migration/tree/main/.cursor/skills/dwr
Command: npx skills add https://github.com/bagrinsergiu/MB-migration --skill dwr

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Orchestrates an end-to-end development workflow by coordinating task decomposition, implementation, and review across isolated subagents in separate contexts, enabling parallel work where possible and robust consolidation of artifacts for progress tracking.

Core Features & Use Cases

  • Multi-agent orchestration: coordinates task-decomposer, task-worker, and work-reviewer in distinct agent contexts to prevent cross-contamination of prompts and contexts.
  • Wave-based parallelism: launches multiple workers in waves to process independent subtasks simultaneously, with strict sequencing and gating between waves.
  • Artifact consolidation: aggregates artifacts from subagents into a centralized session.md, preserving task alignment and progress history.
  • Complexity routing & fallbacks: routes execution via nano, lite, or full complexity branches based on decomposition results and safeguards (BLOCKED, NOT MET, Discoveries).
  • Fix loop workflow: supports mid-cycle decomposition and iterative improvements when reviews reveal critical issues, with guardrails and a 3-round limit.
  • Lasting governance: enforces separation of concerns, with each step performed by its dedicated subagent to ensure reproducibility and auditability.

Quick Start

Initiate the full development cycle by starting the orchestrator to run decomposition, implementation, and review in separate agent contexts.

Frequently Asked Questions about dwr

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

FAQPage Schema
How do I orchestrate multi-agent development workflows with task decomposition and review?

Multi-agent orchestration coordinates task decomposition, implementation, and review across isolated subagents in separate contexts to prevent cross-contamination. It enables parallel work where possible and robust consolidation of artifacts for progress tracking.

What is wave-based parallelism in multi-agent task execution?

Wave-based parallelism launches multiple workers in waves to process independent subtasks simultaneously, with strict sequencing and gating between waves. This approach maximizes throughput while maintaining execution order for dependent tasks.

How do I manage isolated subagent contexts for task-worker and work-reviewer roles?

Isolated subagent contexts are managed by assigning distinct roles—task-decomposer, task-worker, and work-reviewer—to separate agent environments. This separation of concerns ensures reproducibility, auditability, and prevents cross-contamination of prompts.

Can I run parallel subtask implementation using multiple workers in separate agent contexts?

Yes, you can run parallel subtask implementation by launching multiple workers in waves. The system processes independent subtasks simultaneously while enforcing strict sequencing and gating between waves to manage dependencies.

How does artifact consolidation work for session tracking in multi-agent systems?

Artifact consolidation aggregates outputs from subagents into a centralized session.md file. This preserves task alignment and progress history, ensuring all development artifacts are tracked and consolidated for review.

What happens when a work-reviewer finds critical issues during the development cycle?

When critical issues are found, a fix loop workflow triggers mid-cycle decomposition and iterative improvements. The system enforces guardrails and a 3-round limit to prevent infinite loops, routing via BLOCKED, NOT MET, or Discoveries statuses.