/rap-check — RAP Compliance Audit

/rap-check — RAP Compliance Audit

User Input

$ARGUMENTS

Instructions

You are auditing the current R project against the UK Government Reproducible Analytical Pipelines (RAP) maturity levels. There are three cumulative levels: Baseline (5 checks), Silver (9 checks, requires all Baseline), and Gold (4 checks, requires all Silver).

1. Detect R project

First, check whether the current directory is an R project:

• Look for .R files, DESCRIPTION, or .Rproj files.
• If none are found, display:

This directory does not appear to be an R project.

Run `/rap-init` to scaffold a new RAP-compliant R project, or navigate
to an existing R project directory.

Then stop — do not proceed with checks.

2. Determine project name

• If DESCRIPTION exists, extract Package: value.
• Otherwise, use the current directory name.

3. Run Baseline checks (B1–B5)

Run each check using the tools available to you (Read, Grep, Glob, Bash). Record the result as PASS, FAIL, or PARTIAL with a brief detail.

ID	Requirement	How to Check
B1	Code in open-source language	Use Glob to find **/*.R or **/*.r files. PASS if at least one found.
B2	Version controlled with Git	Check if .git/ directory exists.
B3	README with reproduction steps	Check README.md exists. Grep for headings containing "run", "getting started", "reproduce", "steps", or "how to". PASS if file exists and contains at least one such heading. PARTIAL if file exists but lacks reproduction steps.
B4	Code has been peer reviewed	Run git log --oneline --merges and check for merge commits. Alternatively check for CONTRIBUTING.md. PASS if either found. PARTIAL if repo is new with no history.
B5	Code is published openly	Run git remote -v and check for a remote pointing to github.com, gitlab.com, or similar public host.

4. Run Silver checks (S1–S9)

Only evaluate Silver if the user requests it or if all Baseline checks pass.

ID	Requirement	How to Check
S1	Minimal manual intervention	Look for _targets.R, a run_all.R script, or a Makefile. PARTIAL if only numbered scripts exist (e.g. 01_*.R, 02_*.R).
S2	Comprehensive documentation	Grep for roxygen2 comments (#' @) in R files. Check README length (> 50 lines). PASS if both present.
S3	Standard directory structure	Check for R/ directory containing .R files.
S4	Reusable functions	Grep for function( in R/ files. PASS if 3 or more function definitions found.
S5	Coding standards	Check for .lintr config file. Alternatively check for styler in DESCRIPTION or scripts.
S6	Testing framework	Check for tests/testthat/ directory with at least one test-*.R or test_*.R file.
S7	Dependency management	Check for renv.lock file.
S8	Automatic logging	Grep across R files for logger::, log_info, log_warn, log_error, message(, or futile.logger. PASS if found in at least 2 files.
S9	Tidy data format	Grep across R files for tidyverse patterns: dplyr::, tidyr::, %>%, `

5. Run Gold checks (G1–G4)

Only evaluate Gold if all Silver checks pass.

ID	Requirement	How to Check
G1	Fully packaged	Check that DESCRIPTION has a Package: field AND a NAMESPACE file exists.
G2	CI/CD automation	Check for .github/workflows/*.yml files, or .gitlab-ci.yml.
G3	Event/schedule triggers	Read CI config files and check for trigger definitions (on: push, on: schedule, on: pull_request).
G4	Changelog & versioning	Check for NEWS.md or CHANGELOG.md. Check that DESCRIPTION has a Version: field.

6. Calculate achieved level

The levels are cumulative:

• Baseline: All B1–B5 must PASS
• Silver: Baseline achieved AND all S1–S9 must PASS
• Gold: Silver achieved AND all G1–G4 must PASS

A PARTIAL counts as not achieved for level calculation but should be noted.

7. Format and display the report

Use this format:

## RAP Compliance Report

**Project**: <project-name>
**Current Level**: BASELINE ✓/✗ | SILVER ✓/✗ | GOLD ✓/✗

### Baseline (<pass-count>/5)
- ✓/✗ B1: <requirement> (<detail>)
- ✓/✗ B2: <requirement> (<detail>)
- ✓/✗ B3: <requirement> (<detail>)
- ✓/✗ B4: <requirement> (<detail>)
- ✓/✗ B5: <requirement> (<detail>)

### Silver (<pass-count>/9)
- ✓/✗ S1: <requirement> (<detail>)
...

### Gold (<pass-count>/4)   [or "not evaluated — complete Silver first"]
- ✓/✗ G1: <requirement> (<detail>)
...

### Next Steps
<numbered list of actionable steps to reach the next level>

8. Provide actionable guidance

For each FAIL or PARTIAL check, include specific remediation:

Check	Remediation
B2	Run git init to initialise version control
B3	Add reproduction steps under a "How to Run" heading in README.md
B4	Open a pull request and have a colleague review your code
B5	Push to GitHub: git remote add origin <url> then git push
S1	Run /rap-pipeline to configure {targets} automation
S5	Run lintr::use_lintr(type = "tidyverse") to create .lintr config
S6	Run /rap-test to set up testthat framework
S7	Run renv::init() to create renv.lock
S8	Add library(logger) and use log_info() in your pipeline scripts
G1	Run usethis::use_namespace() and ensure DESCRIPTION has a Package field
G2	Add a GitHub Actions workflow: usethis::use_github_action("check-standard")
G4	Run usethis::use_news_md() to create NEWS.md

9. Edge Case Handling

Handle these situations gracefully:

• No R/ directory: If R/ does not exist but .R files exist at the root, still run checks but note in S3 that the standard directory structure is not followed.

• Git not initialised: If .git/ does not exist, B2 FAILS. Guidance: "Run git init or /rap-init to initialise version control."

• packrat instead of renv: If packrat/packrat.lock exists but renv.lock does not, mark S7 as PARTIAL: "Using {packrat} — consider migrating to {renv}."

• Non-standard test directory: If tests/ exists but not tests/testthat/, mark S6 as PARTIAL: "Test directory found but not using {testthat} framework."

• Multiple .Rproj files: Use the first one found and note it.

• Large repositories: If R/ contains more than 50 files, summarise rather than listing every check detail.

10. RAP context

End your response with:

RAP Maturity Levels are defined by the NHS RAP Community of Practice based on the UK Government Analysis Function RAP strategy. Achieving higher levels improves reproducibility, transparency, and efficiency of official statistics production.

Learn more:

• Building RAPs with R
• RAP Companion