pypreset

A meta-tool for scaffolding Python projects with configurable YAML presets.
Supports Poetry, uv, and setuptools, generates CI workflows, testing scaffolds, type checking configs, and more.

mcp-name: io.github.KaiErikNiermann/pypreset

Features

• Preset-based project creation from YAML configs with single inheritance
• Augment existing projects with CI workflows, tests, Docker, documentation, and more
• Three package managers: Poetry, uv (PEP 621 + hatchling), and setuptools (PEP 621 + setuptools.build_meta)
• Two layout styles: src/ layout and flat layout
• Type checking: mypy, pyright, ty, or none
• Code quality: ruff linting/formatting, radon complexity checks, pre-commit hooks
• Docker & devcontainer: generate multi-stage Dockerfiles, .dockerignore, and VS Code devcontainer configs (Docker or Podman)
• Coverage integration: Codecov support with configurable thresholds and ignore patterns
• Documentation scaffolding: MkDocs (Material theme) or Sphinx (RTD theme) with optional GitHub Pages deployment
• Multi-environment testing: tox configuration with tox-uv backend
• pyenv / .python-version: generate .python-version for pyenv and uv, with python-version-file in CI workflows
• Version management: bump-my-version integration, GitHub release automation via gh CLI
• Workflow verification: local GitHub Actions testing with act (auto-detect, auto-install, dry-run and full-run modes)
• PyPI metadata management: read, set, and check publish-readiness of pyproject.toml metadata
• User defaults: persistent config at ~/.config/pypreset/config.yaml
• MCP server: expose all functionality to AI coding assistants via the Model Context Protocol

Installation

pip install pypreset

# With MCP server support
pip install pypreset[mcp]

Quick Start

# Create a CLI tool project with Poetry
pypreset create my-cli --preset cli-tool

# Create a data science project with uv
pypreset create my-analysis --preset data-science --package-manager uv

# Create an empty package with src layout (default)
pypreset create my-package --preset empty-package

# Create a Discord bot
pypreset create my-bot --preset discord-bot

# Create a project with Docker support
pypreset create my-service --preset cli-tool --docker --devcontainer

# Create with .python-version for pyenv/uv
pypreset create my-lib --pyenv --python-version 3.13

# Create with Podman, Codecov, docs, and tox
pypreset create my-project --preset empty-package \
    --container-runtime podman --docker \
    --coverage-tool codecov --coverage-threshold 80 \
    --docs mkdocs --docs-gh-pages \
    --tox

Commands

create -- Scaffold a new project

pypreset create <name> [OPTIONS]

augment -- Add components to an existing project

Analyzes pyproject.toml to auto-detect your tooling, then generates the selected components. Runs in interactive mode by default (prompts for values it can't detect); use --auto to skip prompts.

pypreset augment [path] [OPTIONS]

Available components:

# Interactive mode (prompts for missing values)
pypreset augment ./my-project

# Auto-detect everything, no prompts
pypreset augment --auto

# Generate only specific components
pypreset augment --test-workflow --lint-workflow --gitignore

# Add Docker and devcontainer
pypreset augment --dockerfile --devcontainer

# Add PyPI publish workflow
pypreset augment --pypi-publish

# Add documentation scaffolding
pypreset augment --docs mkdocs

# Generate a README from your project metadata
pypreset augment --readme

# Overwrite existing files
pypreset augment --force

workflow -- Local workflow verification

Verify GitHub Actions workflows locally using act. The proxy auto-detects whether act is installed, can install it on supported systems, and surfaces all act output directly.

# Verify all workflows (dry-run, no containers)
pypreset workflow verify

# Verify a specific workflow file
pypreset workflow verify --workflow .github/workflows/ci.yaml

# Verify a specific job
pypreset workflow verify --job lint

# Full run (executes in containers, requires Docker)
pypreset workflow verify --full-run

# Auto-install act if missing
pypreset workflow verify --auto-install

# Pass extra flags to act
pypreset workflow verify --flag="--secret=GITHUB_TOKEN=xxx"

# Check if act is installed
pypreset workflow check-act

# Install act automatically
pypreset workflow install-act

Supported auto-install targets: Arch Linux (pacman), Ubuntu/Debian (apt), Fedora (dnf), macOS/Linux with Homebrew. Other systems get a link to the act installation page.

version -- Release management

pypreset version release --bump patch     # 0.1.0 -> 0.1.1
pypreset version release --bump minor     # 0.1.0 -> 0.2.0
pypreset version release --bump major     # 0.1.0 -> 1.0.0
pypreset version release-version 2.0.0    # Explicit version
pypreset version rerun <ver>              # Re-tag and push an existing version
pypreset version rerelease <ver>          # Delete and recreate a GitHub release

Requires the gh CLI to be installed and authenticated.

metadata -- PyPI metadata management

pypreset metadata show                                   # Display current metadata
pypreset metadata set --description "My cool package"    # Set description
pypreset metadata set --github-owner myuser              # Auto-generate URLs
pypreset metadata set --license MIT --keyword python     # Set license and keywords
pypreset metadata check                                  # Check publish-readiness

badges -- Generate badge markdown

Reads pyproject.toml to detect your project name, repository URL, and license, then prints badge markdown you can paste into your README.

pypreset badges                  # Badges for current directory
pypreset badges ./my-project     # Badges for a specific project

Other commands

pypreset list-presets              # List all available presets
pypreset show-preset <name>        # Show full preset details
pypreset validate [path]           # Validate project structure
pypreset analyze [path]            # Detect and display project tooling
pypreset config show               # Show current user defaults
pypreset config init               # Create default config file
pypreset config set <key> <value>  # Set a config value

Presets

Built-in presets: empty-package, cli-tool, data-science, discord-bot.

Presets are YAML files that define metadata, dependencies, directory structure, testing, formatting, and more. They support single inheritance via the base: field. Presets can override the README template by setting metadata.readme_template to a custom .j2 filename.

Custom presets

Place custom preset files in ~/.config/pypreset/presets/ or pass a file directly:

pypreset create my-project --config ./my-preset.yaml

User presets take precedence over built-in presets with the same name.

User Configuration

Persistent defaults are stored at ~/.config/pypreset/config.yaml and applied as the lowest-priority layer (presets and CLI flags override them).

pypreset config init                    # Create with defaults
pypreset config set layout flat         # Set default layout
pypreset config set type_checker ty     # Set default type checker
pypreset config show                    # View current config

MCP Server

pypreset is published to the MCP Registry as io.github.KaiErikNiermann/pypreset.

Install via the registry (recommended):

# Claude Code
claude mcp add pypreset -- uvx --from "pypreset[mcp]" pypreset-mcp

# Or add manually to ~/.claude/settings.json

{
  "mcpServers": {
    "pypreset": {
      "command": "uvx",
      "args": ["--from", "pypreset[mcp]", "pypreset-mcp"]
    }
  }
}

Or install locally:

pip install pypreset[mcp]

{
  "mcpServers": {
    "pypreset": {
      "command": "pypreset-mcp",
      "args": []
    }
  }
}

Available tools:

Resources: preset://list, config://user, template://list

Prompts: create-project, augment-project

Development

All tasks use the Justfile:

just install     # Install dependencies
just test        # Run tests
just test-cov    # Tests with coverage
just lint        # Ruff check
just format      # Ruff format
just typecheck   # Pyright
just radon       # Cyclomatic complexity check
just check       # lint + typecheck + radon + test
just all         # format + lint-fix + typecheck + radon + test

See CONTRIBUTING.md for development setup and guidelines.

License

MIT