rapskill

RapSkill

A Claude Code plugin that helps analysts implement Reproducible Analytical Pipelines (RAP) in R and Python, based on the UK Government Analysis Function methodology.

RAP applies software engineering best practices — version control, automated testing, dependency management, and build automation — to the production of official statistics, replacing manual spreadsheet processes with reproducible, auditable code.

Quick Start

Claude Code has a built-in plugin marketplace. No bootstrap or shell commands required.

1. Register this repository

In Claude Code, run:

/plugin marketplace add ivyleavedtoadflax/RapSkill

2. Install the RAP plugin

/plugin install rap

This installs 10 RAP skill files into your project, making them available as slash commands.

3. Create a RAP project

/rap-init my-statistics-pipeline       # R project
/rap-py-init my-statistics-pipeline    # Python project

Available Commands

R Commands

Command	Description
/rap-init [name]	Scaffold a RAP-compliant R project with standard directory structure, configuration files, and starter templates
/rap-check	Audit your R project against the three UK government RAP maturity levels (Baseline, Silver, Gold)
/rap-output <type>	Generate accessible statistical outputs: spreadsheet (a11ytables), charts (GOV.UK theme), report (R Markdown)
/rap-test	Set up testthat infrastructure and generate test skeletons for existing R functions
/rap-pipeline	Configure {targets} build automation with a dependency-aware _targets.R

Python Commands

Command	Description
/rap-py-init [name]	Scaffold a RAP-compliant Python project with src/ layout, pyproject.toml, and uv dependency management
/rap-py-check	Audit your Python project against the three UK government RAP maturity levels (Baseline, Silver, Gold)
/rap-py-output <type>	Generate accessible statistical outputs: spreadsheet (gptables), charts (matplotlib GOV.UK theme), report (Jupyter notebook or Quarto)
/rap-py-test	Set up pytest infrastructure and generate test skeletons for existing Python functions
/rap-py-pipeline [type]	Configure build automation: make (default, Makefile) or dvc (for complex data pipelines)

RAP Maturity Levels

The UK government defines three cumulative levels of RAP compliance. /rap-check and /rap-py-check audit your project against all 18 checks.

Baseline (5 checks)

1. Code written in an open-source language (R or Python)
2. Version controlled with Git
3. Repository includes README with reproduction steps
4. Code has been peer reviewed
5. Code is published openly

Silver (9 additional checks)

Requirement	R Tooling	Python Tooling
Minimal manual intervention	{targets}	Makefile / DVC
Comprehensive documentation	roxygen2	Docstrings (numpy-style)
Standard directory structure	R/ package layout	src/ package layout
Reusable functions	R functions	Python functions
Coding standards	lintr / styler	ruff
Testing framework	testthat	pytest
Dependency management	renv	uv
Automatic logging	{logger}	logging (stdlib)
Tidy data format	tidyverse	pandas

Gold (4 additional checks)

Requirement	R Tooling	Python Tooling
Fully packaged	R package (NAMESPACE)	Python package (pyproject.toml + build-system)
CI/CD automation	GitHub Actions	GitHub Actions
Event/schedule triggers	CI triggers	CI triggers
Changelog & versioning	NEWS.md	CHANGELOG.md

Project Structure