Project Init Skill for Claude

Table of Contents

• Use When
• Workflow
• 1. Detect or Select Language
• 2. Collect Project Metadata
• 3. Review Existing Files
• 4. Render and Apply Templates
• 5. Initialize Git (if needed))
• 6. Verify Setup
• 7. Next Steps
• Error Handling
• Success Criteria
• Examples
• Example 1: New Python Project

Project Initialization Skill

Interactive workflow for initializing new software projects with complete development infrastructure.

Use When

• Starting a new Python, Rust, or TypeScript project
• Updating existing project tooling to current standards
• Need to set up git, GitHub workflows, pre-commit hooks, Makefile
• Want consistent project structure across team
• Converting unstructured project to best practices
• Adding missing configurations to established codebases

Workflow

1. Detect or Select Language

Load modules/language-detection.md

• Auto-detect from existing files (pyproject.toml, Cargo.toml, package.json)
• If ambiguous or empty directory, ask user to select
• Validate language is supported (python, rust, typescript)

2. Collect Project Metadata

Load modules/metadata-collection.md

Gather:

• Project name (default: directory name)
• Author name and email
• Project description
• Language-specific settings:
• Python: version (default 3.10)
• Rust: edition (default 2021)
• TypeScript: framework (React, Vue, etc.)
• License type (MIT, Apache, GPL, etc.)

3. Review Existing Files

Check for existing configurations:

ls -la

Verification: Run the command with --help flag to verify availability.

If files exist (Makefile, .gitignore, etc.):

• Show what would be overwritten
• Ask for confirmation or selective overwrite
• Offer merge mode (preserve custom content)

4. Render and Apply Templates

Load modules/template-rendering.md

Run initialization script:

python3 plugins/attune/scripts/attune_init.py \
  --lang {{LANGUAGE}} \
  --name {{PROJECT_NAME}} \
  --author {{AUTHOR}} \
  --email {{EMAIL}} \
  --python-version {{PYTHON_VERSION}} \
  --description {{DESCRIPTION}} \
  --path .

Verification: Run the command with --help flag to verify availability.

The script also scaffolds the project decision journal: docs/tradeoffs.md

and docs/lessons-learned.md. These are append-only logs that later workflows

(brainstorm, specify, plan, execute, review) write to as decisions and lessons

arise. Existing journal files are never overwritten. The format follows the

leyline:decision-journal contract; init uses leyline's template when present

and a vendored copy otherwise, so it works with or without leyline installed.

Verification: Confirm docs/tradeoffs.md and docs/lessons-learned.md

exist and each contains an ## Active index and an ## Archive section.

5. Initialize Git (if needed)

# Check if git is initialized
if [ ! -d .git ]; then
  git init
  echo "Git repository initialized"
fi

Verification: Run git status to confirm working tree state.

6. Verify Setup

Validate setup:

# Check Makefile targets
make help

# List created files
git status

Verification: Run git status to confirm working tree state.

7. Next Steps

Advise user to:

# Install dependencies and hooks
make dev-setup

# Run tests to verify setup
make test

# See all available commands
make help

Verification: Run pytest -v to verify tests pass.

Error Handling

• Language detection fails: Ask user to specify --lang
• Script not found: Guide to plugin installation location
• Permission denied: Suggest chmod +x on scripts
• Git conflicts: Offer to stash or commit existing work

Success Criteria

• All template files created successfully
• No overwrites without user confirmation
• Git repository initialized
• make help shows available targets
• make test runs without errors (even if no tests yet)

Exit Criteria

• [ ] Template files for the selected language are created (or, in dry-run,

reported) with no unconfirmed overwrites.

• [ ] docs/tradeoffs.md and docs/lessons-learned.md exist, each with an

## Active index and an ## Archive section.

• [ ] A pre-existing journal file is detected and left unmodified.
• [ ] Git is initialized (unless --no-git) and make help lists targets.
• [ ] Running with --dry-run writes no files, including the journal.

Examples

Example 1: New Python Project

**Verification:** Run `pytest -v` to verify tests pass.
User: /attune:project-init