Back to Skills

submit-to-cran

pjt222
Updated Yesterday
1 views
17
2
17
View on GitHub
Metadesign

About

This Claude Skill provides a complete workflow for submitting R packages to CRAN, covering both initial releases and updates. It automates pre-submission checks on local, win-builder, and R-hub platforms, along with preparing the required `cran-comments.md` file. Developers should use it when a package is ready for its first CRAN release, for submitting updates, or when re-submitting after reviewer feedback.

Quick Install

Claude Code

Recommended
Primary
npx skills add pjt222/agent-almanac -a claude-code
Plugin CommandAlternative
/plugin add https://github.com/pjt222/agent-almanac
Git CloneAlternative
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/submit-to-cran

Copy and paste this command in Claude Code to install this skill

Documentation

Submit to CRAN

Execute the full CRAN submission workflow from pre-flight checks through submission.

When to Use

  • Package is ready for initial CRAN release
  • Submitting an updated version of an existing CRAN package
  • Re-submitting after CRAN reviewer feedback

Inputs

  • Required: R package passing local R CMD check with 0 errors and 0 warnings
  • Required: Updated version number in DESCRIPTION
  • Required: Updated NEWS.md with changes for this version
  • Optional: Previous CRAN reviewer comments (for re-submissions)

Procedure

Step 1: Version and NEWS Check

Verify DESCRIPTION has the correct version:

desc::desc_get_version()

Verify NEWS.md has an entry for this version. The entry should summarize user-facing changes.

Got: Version follows semantic versioning. NEWS.md has a matching entry for this version.

If fail: Update version with usethis::use_version() (choose "major", "minor", or "patch"). Add a NEWS.md entry summarizing user-facing changes.

Step 2: Local R CMD Check

devtools::check()

Got: 0 errors, 0 warnings, 0 notes (1 note acceptable for new submissions: "New submission").

If fail: Fix all errors and warnings before proceeding. Read the check log at <pkg>.Rcheck/00check.log for details. Notes should be explained in cran-comments.md.

Step 3: Spell Check

devtools::spell_check()

Add legitimate words to inst/WORDLIST (one word per line, sorted alphabetically).

Got: No unexpected misspellings. All flagged words are either corrected or added to inst/WORDLIST.

If fail: Fix genuine misspellings. For legitimate technical terms, add them to inst/WORDLIST (one word per line, alphabetically sorted).

Step 4: URL Check

urlchecker::url_check()

Got: All URLs return HTTP 200. No broken or redirected links.

If fail: Replace broken URLs. Use \doi{} for DOI links instead of raw URLs. Remove links to resources that no longer exist.

Step 5: Win-Builder Checks

devtools::check_win_devel()
devtools::check_win_release()

Wait for email results (usually 15-30 minutes).

Got: 0 errors, 0 warnings on both Win-builder release and devel. Results arrive by email within 15-30 minutes.

If fail: Address platform-specific issues. Common causes: different compiler warnings, missing system dependencies, path separator differences. Fix locally and re-submit to Win-builder.

Step 6: R-hub Check

rhub::rhub_check()

This checks on multiple platforms (Ubuntu, Windows, macOS).

Got: All platforms pass with 0 errors and 0 warnings.

If fail: If a specific platform fails, check the R-hub build log for platform-specific errors. Use testthat::skip_on_os() or conditional code for platform-dependent behavior.

Step 7: Prepare cran-comments.md

Create or update cran-comments.md in the package root:

## R CMD check results
0 errors | 0 warnings | 1 note

* This is a new release.

## Test environments
* local: Windows 11, R 4.5.0
* win-builder: R-release, R-devel
* R-hub: ubuntu-latest (R-release), windows-latest (R-release), macos-latest (R-release)

## Downstream dependencies
There are currently no downstream dependencies for this package.

For updates, include:

  • What changed (brief)
  • Response to any previous reviewer feedback
  • Reverse dependency check results if applicable

Got: cran-comments.md accurately summarizes check results across all test environments and explains any notes.

If fail: If check results differ across platforms, document all variations. CRAN reviewers will check these claims against their own tests.

Step 8: Final Pre-flight

# One last check
devtools::check()

# Verify the built tarball
devtools::build()

Got: Final devtools::check() passes cleanly. A .tar.gz tarball is built in the parent directory.

If fail: If a last-minute issue appears, fix it and re-run all checks from Step 2. Do not submit with known failures.

Step 9: Submit

devtools::release()

This runs interactive checks and submits. Answer all questions honestly.

Alternatively, submit manually at https://cran.r-project.org/submit.html by uploading the tarball.

Got: Confirmation email from CRAN arrives within minutes. Click the confirmation link to finalize the submission.

If fail: Check email for rejection reasons. Common issues: examples too slow, missing \value tags, non-portable code. Fix the issues and re-submit, noting in cran-comments.md what changed.

Step 10: Post-Submission

After acceptance:

# Tag the release
usethis::use_github_release()

# Bump to development version
usethis::use_dev_version()

Got: GitHub release is created with the accepted version tag. DESCRIPTION is bumped to the development version (x.y.z.9000).

If fail: If the GitHub release fails, create it manually with gh release create. If CRAN acceptance is delayed, wait for the confirmation email before tagging.

Validation

  • R CMD check returns 0 errors, 0 warnings on local machine
  • Win-builder passes (release + devel)
  • R-hub passes on all tested platforms
  • cran-comments.md accurately describes check results
  • All URLs valid
  • No spelling errors
  • Version number is correct and incremented
  • NEWS.md is current
  • DESCRIPTION metadata is complete and accurate

Pitfalls

  • Examples too slow: Wrap expensive examples in \donttest{}. CRAN enforces time limits.
  • Non-standard file/directory names: Avoid files that trigger CRAN notes (check .Rbuildignore)
  • Missing \value in docs: All exported functions need a @return tag
  • Vignette build failures: Ensure vignettes build in a clean environment without your .Renviron
  • DESCRIPTION Title format: Must be Title Case, no period at end, no "A Package for..."
  • Forgetting reverse dependency checks: For updates, run revdepcheck::revdep_check()

Examples

# Full pre-submission workflow
devtools::spell_check()
urlchecker::url_check()
devtools::check()
devtools::check_win_devel()
rhub::rhub_check()
# Wait for results...
devtools::release()

Related Skills

  • release-package-version - version bumping and git tagging
  • write-roxygen-docs - ensure documentation meets CRAN standards
  • setup-github-actions-ci - CI checks that mirror CRAN expectations
  • build-pkgdown-site - documentation site for accepted packages

GitHub Repository

pjt222/agent-almanac
Path: i18n/caveman-lite/skills/submit-to-cran
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

Related Skills

content-collections

Meta

This skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.

View skill

polymarket

Meta

This skill enables developers to build applications with the Polymarket prediction markets platform, including API integration for trading and market data. It also provides real-time data streaming via WebSocket to monitor live trades and market activity. Use it for implementing trading strategies or creating tools that process live market updates.

View skill

creating-opencode-plugins

Meta

This skill helps developers create OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It provides the plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript modules. Use it when you need to intercept, monitor, or extend the OpenCode AI assistant's lifecycle with custom event-driven logic.

View skill

sglang

Meta

SGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.

View skill