What happens if target files already exist?

READMe, CHANGELOG, and LICENSE are skipped to preserve content; .gitignore is merged; _NOTES preserved; AGENTS is created or augmented as needed.

Which project types are supported?

Supported types include general, code-python, and code-node, with type-specific additions like src/, tests/, and package configuration.

How are templates customized?

Templates are populated from assets using placeholders {{PROJECT_NAME}}, {{DESCRIPTION}}, {{DATE}}, and {{YEAR}} to tailor each project outgoing document.

init-project

Scanned

npx machina-cli add skill product-on-purpose/pm-skills/init-project --openclaw

Files (1)

SKILL.md

3.2 KB

Project Initialization Workflow

Scaffold projects with agentic coding structure for AI-assisted development.

Execution Steps

1. Determine Target Directory

Default: Current working directory
If user specifies a path, use that instead
If directory doesn't exist, offer to create it

2. Gather Requirements

Required	Question	Default
	Project name?	Directory name
	Project type?	`general`
	License preference?	MIT
	Brief description?	"A new project"

Skip questions if context provides answers.

3. Check for Existing Files

Before creating, check what already exists:

If Exists	Action
README.md	Skip — preserve existing
CHANGELOG.md	Skip — preserve existing
LICENSE	Skip — preserve existing
.gitignore	Merge — append missing entries
_NOTES/	Skip — preserve existing
AGENTS/	Create missing parts only

This allows safe re-runs on existing projects to add agentic structure.

4. Confirm Before Creating

Show user:

Target path
Files to be created (noting any skipped)
Project type selected

Wait for confirmation.

5. Create Directory Structure

<project-root>/
├── README.md
├── CHANGELOG.md
├── LICENSE
├── .gitignore
├── _NOTES/
│   └── .gitkeep
└── AGENTS/
    └── claude-opus-4.5/
        ├── CONTEXT.md
        ├── TODO.md
        ├── DECISIONS.md
        └── SESSION-LOG/

6. Populate Files

Use templates from assets/ folder, substituting:

{{PROJECT_NAME}} — Project name
{{DESCRIPTION}} — Project description
{{DATE}} — Current date (YYYY-MM-DD)
{{YEAR}} — Current year

7. Add Type-Specific Files

Type	Additional Structure
`general`	Base structure only
`code-python`	+ `src/`, `tests/`, `pyproject.toml`
`code-node`	+ `src/`, `package.json`

See references/project-types.md for details.

8. Confirm Completion

Report:

Full path created
Files generated
Suggested next steps

Template Assets

File	Source
README.md	`assets/README.template.md`
CHANGELOG.md	`assets/CHANGELOG.template.md`
LICENSE (MIT)	`assets/LICENSE-MIT.txt`
LICENSE (Apache)	`assets/LICENSE-Apache2.txt`
.gitignore	`assets/gitignore-general.txt`
CONTEXT.md	`assets/CONTEXT.template.md`
TODO.md	`assets/TODO.template.md`
DECISIONS.md	`assets/DECISIONS.template.md`

Integration

This skill creates structure compatible with /wrap-session:

Init Creates	Wrap-Session Updates
README.md	README.md (with progress)
CHANGELOG.md	CHANGELOG.md (with changes)
CONTEXT.md	CONTEXT.md (current state)
TODO.md	TODO.md (with tasks)
SESSION-LOG/	SESSION-LOG/*.md

Source

git clone https://github.com/product-on-purpose/pm-skills/blob/main/.claude/skills/init-project/SKILL.mdView on GitHub

Overview

Init-project creates a standardized AGENTS folder, documentation templates, and a _NOTES scratch space for new or existing projects. It supports multiple project types and safe re-runs by preserving or merging existing files.

How This Skill Works

It first selects the target directory, collects project requirements, and audits existing files, then creates the AGENTS folder, _NOTES, and base templates. It populates templates from assets with placeholders like {{PROJECT_NAME}}, {{DESCRIPTION}}, {{DATE}}, and {{YEAR}}, and adds type-specific structures (general, code-python, code-node). It remains compatible with wrap-session updates for incremental improvements.

When to Use It

Starting a brand-new project that requires an agentic structure
Adding AI agent support to an existing codebase
When the user says init, initialize, setup project, or scaffold
Re-running the scaffold to progressively augment missing AGENTS content
Setting up code-python or code-node project templates

Quick Start

Step 1: Determine target directory (default to current working dir or user-specified path)
Step 2: Gather requirements and choose project type (general, code-python, code-node)
Step 3: Create directory structure, populate templates, and review before finalizing

Best Practices

Clarify the target path before creating to avoid misplacement
Review the existing files to determine what should be preserved or merged
Use the assets templates for consistent documentation and structure
Leverage the _NOTES scratch space for quick, interim notes during setup
Test re-runs in a safe environment to validate incremental improvements

Example Use Cases

Bootstrapping a new AI-assisted Python project with a dedicated AGENTS folder and NOTES space
Augmenting an existing repo to add agentic structure without overwriting README or CHANGELOG
Scaffolding a Node.js project to include src/, tests/, and a ready-to-use AGENTS/claude-opus-4.5/ layout
Migrating a simple project to an agentic setup while preserving licenses and docs
Setting up a general project with base templates and wrap-session compatible outputs

Frequently Asked Questions

Add this skill to your agents