返回技能列表

submit-to-cran

pjt222
更新于 2 days ago
6 次查看
17
2
17
在 GitHub 上查看
design

关于

This Claude Skill automates the complete workflow for submitting an R package to CRAN, including pre-submission checks and preparing the required `cran-comments.md` file. It handles initial releases, updates to existing packages, and re-submissions after reviewer feedback. Developers should use it when their package passes local checks with zero errors and warnings and has an updated version number.

快速安装

Claude Code

推荐
主要方式
npx skills add pjt222/agent-almanac -a claude-code
插件命令备选方式
/plugin add https://github.com/pjt222/agent-almanac
Git 克隆备选方式
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/submit-to-cran

在 Claude Code 中复制并粘贴此命令以安装该技能

技能文档

Submit to CRAN

Full CRAN sub workflow: pre-flight checks → submission.

Use When

  • Pkg ready for initial CRAN release
  • Sub updated ver of existing CRAN pkg
  • Re-sub after CRAN reviewer feedback

In

  • Required: R pkg passing local R CMD check w/ 0 err + 0 warn
  • Required: Updated ver # in DESCRIPTION
  • Required: Updated NEWS.md w/ ver changes
  • Optional: Prev CRAN reviewer comments (re-subs)

Do

Step 1: Ver + NEWS Check

Verify DESCRIPTION ver:

desc::desc_get_version()

Verify NEWS.md has entry. Summarize user-facing changes.

Got: Ver follows semver. NEWS.md has matching entry.

If err: Update ver usethis::use_version() (major/minor/patch). Add NEWS.md entry.

Step 2: Local R CMD Check

devtools::check()

Got: 0 err, 0 warn, 0 notes (1 note OK new sub: "New submission").

If err: Fix all err+warn before. Read log <pkg>.Rcheck/00check.log. Notes → explain in cran-comments.md.

Step 3: Spell Check

devtools::spell_check()

Add legit words → inst/WORDLIST (one per line, sorted).

Got: No unexpected misspellings. All flagged corrected | added.

If err: Fix genuine misspellings. Tech terms → inst/WORDLIST sorted.

Step 4: URL Check

urlchecker::url_check()

Got: All URLs HTTP 200. No broken/redirected.

If err: Replace broken. \doi{} for DOI links not raw URLs. Remove dead links.

Step 5: Win-Builder

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

Wait email (~15-30 min).

Got: 0 err + 0 warn on both release + devel. Email in 15-30 min.

If err: Address platform-specific. Common: diff compiler warns, missing sys deps, path sep diffs. Fix local + re-sub.

Step 6: R-hub Check

rhub::rhub_check()

Multi-platform (Ubuntu, Windows, macOS).

Got: All platforms pass 0 err + 0 warn.

If err: Specific platform fails → check R-hub log. Use testthat::skip_on_os() | conditional code for platform-dep behavior.

Step 7: Prep cran-comments.md

Create | update in pkg 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.

Updates → include:

  • What changed (brief)
  • Response to prev reviewer feedback
  • Reverse dep check results if applicable

Got: Accurate summary across all envs, explains notes.

If err: Results differ across platforms → doc all variations. CRAN reviewers check vs own tests.

Step 8: Final Pre-flight

# One last check
devtools::check()

# Verify the built tarball
devtools::build()

Got: Final check passes clean. .tar.gz built in parent dir.

If err: Last-min issue → fix + re-run all from Step 2. Don't sub w/ known fails.

Step 9: Submit

devtools::release()

Interactive checks + sub. Answer honest.

Alt: manual at https://cran.r-project.org/submit.html, upload tarball.

Got: Confirmation email from CRAN in min. Click link → finalize.

If err: Check email for rejection reasons. Common: examples too slow, missing \value tags, non-portable code. Fix + re-sub, note in cran-comments.md what changed.

Step 10: Post-Submission

Post-acceptance:

# Tag the release
usethis::use_github_release()

# Bump to development version
usethis::use_dev_version()

Got: GitHub release created w/ accepted ver tag. DESCRIPTION bumped to dev (x.y.z.9000).

If err: GH release fails → manual gh release create. CRAN acceptance delayed → wait email before tag.

Check

  • R CMD check 0 err + 0 warn local
  • Win-builder passes (release + devel)
  • R-hub passes all platforms
  • cran-comments.md accurate
  • All URLs valid
  • No spelling errors
  • Ver # correct + incremented
  • NEWS.md current
  • DESCRIPTION metadata complete + accurate

Traps

  • Examples too slow: Wrap expensive in \donttest{}. CRAN enforces time limits.
  • Non-std file/dir names: Avoid files triggering CRAN notes (check .Rbuildignore)
  • Missing \value in docs: All exported fns need @return tag
  • Vignette build fails: Vignettes must build in clean env w/o your .Renviron
  • DESCRIPTION Title: Title Case, no period at end, no "A Package for..."
  • Forget rev dep checks: 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()

  • release-package-version — ver bumping + git tagging
  • write-roxygen-docs — docs meet CRAN standards
  • setup-github-actions-ci — CI mirroring CRAN expectations
  • build-pkgdown-site — docs site for accepted pkgs

GitHub 仓库

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

相关推荐技能

content-collections

Content Collections 是一个 TypeScript 优先的构建工具,可将本地 Markdown/MDX 文件转换为类型安全的数据集合。它专为构建博客、文档站和内容密集型 Vite+React 应用而设计,提供基于 Zod 的自动模式验证。该工具涵盖从 Vite 插件配置、MDX 编译到生产环境部署的完整工作流。

查看技能

polymarket

这个Claude Skill为开发者提供完整的Polymarket预测市场开发支持,涵盖API调用、交易执行和市场数据分析。关键特性包括实时WebSocket数据流,可监控实时交易、订单和市场动态。开发者可用它构建预测市场应用、实施交易策略并集成实时市场预测功能。

查看技能

creating-opencode-plugins

该Skill帮助开发者创建OpenCode插件,用于接入命令、文件、LSP等25+种事件。它提供了插件结构、事件API规范和JavaScript/TypeScript实现模式,适合需要拦截操作、扩展功能或自定义事件处理的场景。开发者可通过它快速构建响应式模块来增强OpenCode AI助手的能力。

查看技能

sglang

SGLang是一个专为LLM设计的高性能推理框架,特别适用于需要结构化输出的场景。它通过RadixAttention前缀缓存技术,在处理JSON、正则表达式、工具调用等具有重复前缀的复杂工作流时,能实现极速生成。如果你正在构建智能体或多轮对话系统,并追求远超vLLM的推理性能,SGLang是理想选择。

查看技能