Best use case
parallel-setup is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
One-time setup of parallel/ directory for multi-agent development
Teams using parallel-setup should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/parallel-setup/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How parallel-setup Compares
| Feature / Agent | parallel-setup | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
One-time setup of parallel/ directory for multi-agent development
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# parallel-setup
**Category**: Parallel Development
## Usage
```bash
/parallel-setup [--tech django|typescript|go]
```
## Arguments
- `--tech`: Optional - Technology stack hint for future decompositions (default: auto-detect)
## Purpose
**One-time** initialization of the `parallel/` directory at project root for parallel multi-agent development. This creates the infrastructure for organizing decomposed tasks by Tech Spec.
> Run this once per project. Then use `/parallel-decompose` for each Tech Spec.
## Execution Instructions for Claude Code
When this command is run, Claude Code should:
### 1. Create Directory Structure
```bash
mkdir -p parallel
```
Creates:
```
parallel/
├── README.md # Explains parallel development workflow
└── .gitignore # What to track vs ignore
```
### 2. Create README.md
Create `parallel/README.md`:
```markdown
# Parallel Development Artifacts
This directory contains artifacts for parallel multi-agent development.
## Structure
Each Tech Spec decomposition creates a subdirectory:
```
parallel/
├── TS-0042-inventory-system/ # Keyed by Tech Spec ID
│ ├── manifest.json # Regeneration metadata
│ ├── context.md # Shared project context
│ ├── tasks/ # Task specifications
│ ├── contracts/ # Shared types & API schema
│ ├── prompts/ # Agent launch prompts
│ ├── scripts/ # Launch & monitor scripts
│ ├── architecture.md # System design
│ └── task-graph.md # Dependency visualization
└── ...
```
## Commands
- `/parallel-decompose <prd> --tech-spec <ts-file>` - Decompose PRD into tasks
- `/parallel-integrate --parallel-dir <dir>` - Verify integration
## Regeneration
Each subdirectory contains a `manifest.json` with metadata to regenerate artifacts:
```bash
# Re-run decomposition with same args
/parallel-decompose docs/prd.md --tech-spec tech-specs/approved/TS-XXXX.md
```
```
### 3. Create .gitignore
Create `parallel/.gitignore`:
```gitignore
# Generated prompts and scripts (regenerate with /parallel-decompose)
*/prompts/
*/scripts/
# Integration reports
*/integration-report.md
# Keep tracked:
# - manifest.json (regeneration metadata)
# - context.md (shared context)
# - tasks/ (task specifications)
# - contracts/ (shared types & API)
# - architecture.md (system design)
# - task-graph.md (dependencies)
```
### 4. Update CLAUDE.md (if exists)
If `CLAUDE.md` exists at project root, add a section:
```markdown
## Parallel Development
This project uses parallel multi-agent development. Artifacts are in `parallel/`.
- Each Tech Spec decomposition creates `parallel/TS-XXXX-{slug}/`
- Task files use compact YAML format for token efficiency
- Contracts are shared via `contracts/` subdirectory
- Run `/parallel-decompose --help` for usage
```
### 5. Report Results
Output summary:
```
Created parallel/ directory structure
parallel/
├── README.md # Workflow documentation
└── .gitignore # Track config
Next steps:
1. Create a Tech Spec: /create-tech-spec <title>
2. Decompose PRD: /parallel-decompose <prd> --tech-spec <ts-file>
```
## Example
```bash
# Initialize parallel development
/parallel-setup
# With technology hint
/parallel-setup --tech django
```
## Notes
- This command is idempotent - safe to run multiple times
- Existing files are NOT overwritten
- Creates foundation for `/parallel-decompose` command
- Each decomposition creates its own subdirectory (e.g., `parallel/TS-0042-slug/`)Related Skills
setup-mcp-auth
Configure authentication for an existing FastMCP server
parallel-ready-django
Audit and prepare a Django codebase for parallel multi-agent development. Use when asked to check if a Django project is ready for parallelization, prepare a repo for multi-agent work, audit codebase structure, set up orchestration infrastructure, or identify blockers for parallel development. Analyzes Django apps, models, migrations, and module boundaries.
parallel-fix-django
Fix Django-specific blockers identified in parallelization readiness assessment
mypy-setup
Set up Mypy type checking configuration for a Python project
django-project-setup
Set up a new Django 6.0 project with modern tooling (uv, direnv, HTMX, OAuth, DRF, testing). Use when the user wants to create a Django project from scratch with production-ready configuration.
parallel-validate-prompts
Validate and fix parallel prompts for required sections
parallel-task-format
Compact YAML format for defining parallel task specifications with scope, boundaries, and agent assignments. Use when creating task files for parallel development.
parallel-run
Orchestrate parallel agent execution with git worktrees
parallel-prompt-generator
Generate agent-ready prompts from existing task specification files. Use when regenerating prompts after editing tasks, updating prompt templates, or preparing tasks for cpo execution.
parallel-integrate
Verify integration after parallel agent execution and generate report
parallel-execution
Execute multiple Claude Code agents in parallel using the cpo CLI tool. Use when running parallel tasks, monitoring execution, or understanding the execution workflow.
parallel-decompose
Decompose PRDs and Tech Specs into parallel-executable tasks with contracts, prompts, and dependency graphs. Use when breaking down a PRD for multi-agent execution.