skills.rest
jpoutrin
jpoutrinCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

parallel-agents

Orchestrates parallel Codex and Gemini agents for PRD-driven software development.

14|1|Updated Dec 5, 2025
Search Tags:#devops#project management#multi-agent#parallel development#git worktrees#ai engineering#task orchestration#prd decomposition
One-click install
npx skills add https://github.com/jpoutrin/product-forge --skill parallel-agents
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: parallel-agents
Source: https://github.com/jpoutrin/product-forge/tree/main/plugins/product-design/skills/parallel-agents
Command: npx skills add https://github.com/jpoutrin/product-forge --skill parallel-agents

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solves?

Orchestrating large development efforts by decomposing Product Requirements Documents (PRDs) into independent, parallel tasks for multiple AI agents is complex and requires careful coordination to maximize efficiency and minimize conflicts. This Skill automates this orchestration.

Core Features & Use Cases

  • Workflow Overview: A structured process from setup (/parallel-setup) to decomposition (/parallel-decompose), execution (/parallel-run), and integration (/parallel-integrate).
  • Decomposition Principles: Guidance on contract-first development, maximizing task independence, explicit boundaries, and optimal task granularity (2-4 hours).
  • Task Spec Format: A compact YAML format for defining individual tasks, including scope, requirements, dependencies, and recommended agent types.
  • Parallel Execution: Manages the creation of git worktrees, launches agents in parallel (respecting wave dependencies), and monitors progress.
  • Integration Checks: Verifies contract compliance, boundary adherence, and overall implementation against the Tech Spec after execution.
  • Use Case: Decomposing a new feature PRD into tasks for parallel AI development, setting up a project for multi-agent work, executing a wave of parallel tasks, or integrating code from multiple agent workstreams.

Quick Start

Use the parallel-agents skill to generate a compact YAML task specification for a 'User Management' task, including scope, requirements, and checklist.

Frequently Asked Questions about parallel-agents

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

FAQPage Schema
How do I decompose a PRD into parallel tasks for multiple AI agents?▼

Decompose a PRD into parallel tasks by using contract-first development principles to define independent work units with explicit boundaries. The parallel-agents skill breaks down requirements into 2-4 hour tasks specified in YAML format, assigns them to agents, and manages execution across isolated git worktrees to maximize concurrency and minimize conflicts.

What's the best way to coordinate multiple AI agents working on the same project simultaneously?▼

Coordinate multiple agents using git worktree isolation for each agent's workstream, standardized task manifests defining scope and dependencies, and wave-based parallel execution that respects task dependencies. The skill orchestrates agent launch, monitors progress, and integrates results while enforcing contract compliance across boundaries.

Can I use git worktrees to isolate parallel development work from multiple agents?▼

Yes. Git worktree isolation is core to parallel-agents; each agent task runs in its own worktree with an isolated artifact directory. This prevents conflicts, allows concurrent modifications, and enables independent code verification before integration.

How do I validate that parallel agent work meets technical requirements before merging?▼

After parallel execution completes, run integration checks to verify contract compliance, boundary adherence, and implementation against the tech spec. The skill generates an integration report that confirms all task outputs satisfy functional and technical requirements before code merges.

What format do I use to specify tasks for parallel AI agent execution?▼

Tasks are specified in a compact YAML format that includes scope, functional and technical requirements, dependencies, checklist items, and recommended agent type. This standardized format ensures clarity, enables dependency ordering, and drives orchestration of parallel waves.

Do I need to manually manage dependencies between parallel agent tasks?▼

No. Define task dependencies explicitly in the YAML task spec; the orchestrator respects wave dependencies and sequences parallel execution automatically. Tasks with no dependencies launch together; dependent tasks wait until prerequisites complete.

Related Skills

Explore More Software Engineering Skills→
konglong87konglong87
46

pm-feature

Decompose product requirements into detailed development specifications from PRD and prototype inputs.

Community
Advanced
saw4405saw4405

worktree-preview

Merge feature branches into a preview worktree for integrated testing.

Community
Advanced
xxihxxih
1

dmux-workflows

Orchestrate parallel AI agent sessions across multiple tools with dmux.

Community
Advanced
patrobpatrob

parallel-workstreams

Launch parallel workstreams in isolated git worktrees with TDD cycles.

Community
Advanced