Do I need a Tech Spec to use this skill?

While a Tech Spec is recommended for better contract definitions, the skill can generate contracts and tasks directly from a PRD with a warning to the user.

Is it specific to a particular programming language?

While the default patterns include Python and React, the underlying logic of breaking down specs into parallel tasks is applicable to any modern software stack.

What is the benefit of the contract-first approach in this skill?

It defines interfaces (API schemas and types) before implementation starts, which prevents integration issues when multiple agents or developers work on related tasks simultaneously.

How does it handle task dependencies?

It creates a Mermaid-based task graph and organizes tasks into 'waves' within a manifest, ensuring prerequisite tasks are completed before dependent ones begin.

What is the primary output of the Parallel Decompose skill?

It generates a structured directory containing a manifest.json, shared context, architecture diagrams, API contracts, and individual task files ready for multi-agent execution.

Parallel Task Decomposer

Name: Parallel Task Decomposer
Author: jpoutrin

byjpoutrin

•

生产力与工作流

Breaks down PRDs and technical specifications into discrete, parallel-executable tasks for multi-agent development workflows.

The Parallel Decompose skill automates the transition from high-level product requirements to actionable development tasks. It analyzes PRDs and Tech Specs to generate a comprehensive execution environment, including shared context, architecture diagrams, data contracts, and dependency graphs. By enforcing a contract-first approach and maximizing task independence, it enables teams and AI agents to work concurrently on complex features without overlapping, significantly reducing development cycles and coordination overhead.

主要功能

01Visual dependency mapping using Mermaid architecture and task graphs

02Contract-first generation including Python types and OpenAPI schemas

03Multi-agent manifest creation with wave-based execution logic

04Standardized task decomposition with explicit boundaries and granularity control

05Automated PRD and Technical Specification analysis and validation

062 GitHub stars

使用场景

01Breaking down a complex feature PRD for a multi-agent CPO execution run

02Visualizing task dependencies and critical paths for complex software architecture changes

03Generating API contracts and data models to enable parallel frontend and backend development

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jpoutrin/product-forge parallel-decompose

For use in Claude.ai and ChatGPT

Download Skill