跳到正文
FunCoding

搜索

搜索文档、Skill 和 MCP

r-cran-extrachecks

Prepare R packages for CRAN submission by checking for common ad-hoc requirements not caught by devtools::check(). Use when: (1) Preparing a package for first CRAN release, (2) Preparing a package update for CRAN resubmission, (3) Reviewing a package to ensure CRAN compliance, (4) Responding to CRAN reviewer feedback. Covers documentation requirements, DESCRIPTION field standards, URL validation, examples, and administrative requirements.

AI 与智能体531r-lib/r-cran-extrachecks/SKILL.md

安装

把这段话发给 Claude Code、Codex 或 Cursor。智能体会先检查安全性,你确认后才安装。

读取 https://funcoding.ai/skills/posit-dev/skills/r-cran-extrachecks/install.md ,按里面的步骤帮我安装这个 Skill。

SKILL.md

CRAN Extra Checks

Help R package developers prepare packages for CRAN submission by systematically checking for common ad-hoc requirements that CRAN reviewers enforce but devtools::check() doesn't catch.

Workflow

  1. Initial Assessment: Ask user if this is first submission or resubmission
  2. Run Standard Checklist: Work through each item systematically (see below)
  3. Identify Issues: As you review files, note specific problems
  4. Propose Fixes: Suggest specific changes for each issue found
  5. Implement Changes: Make edits only when user approves
  6. Verify: Confirm all changes are complete

Standard CRAN Preparation Checklist

Work through these items systematically:

  1. Create NEWS.md: Run usethis::use_news_md() if not already present
  2. Create cran-comments.md: Run usethis::use_cran_comments() if not already present
  3. Review README:
    • Ensure it includes install instructions that will be valid when the package is accepted to CRAN (usually install.packages("pkgname")).
    • Check that it does not contain relative links. This works on GitHub but will be flagged by CRAN. Use full URLs to package documentation or remove the links.
    • Does the README clearly explain the package purpose and functionality?
    • Important: If README.Rmd exists, edit ONLY README.Rmd (README.md will be overwritten), then run devtools::build_readme() to re-render README.md
  4. Proofread DESCRIPTION: Carefully review Title: and Description: fields (see detailed guidance below)
  5. Check function documentation: Verify all exported functions have @return and @examples (see detailed guidance below). Note the exemption for S3 methods documented only with #' @export: they produce no .Rd file and are not required to have either tag
  6. Verify copyright holder: Check that Authors@R: includes a copyright holder with role [cph]
  7. Review bundled file licensing: Check licensing of any included third-party files
  8. Run URL checks: Use urlchecker::url_check() and fix any issues

Detailed CRAN Checks

Documentation Requirements

Return Value Documentation (Strictly Enforced)

CRAN now strictly requires @return documentation for all exported functions. Use the roxygen2 tag @return to document what the function returns.

  • Required even for functions marked @keywords internal
  • Required even if function returns nothing - document as @return None or similar
  • Must be present for every exported function

These rules apply only to functions with an .Rd file. Roxygen2 creates an .Rd file only when a block contains documentation content (a title or description). A block that contains nothing but #' @export adds the function to the NAMESPACE without generating an .Rd file, so there is nothing for CRAN's documentation checks to inspect. This is the conventional pattern for S3 methods:

#' @export
print.my_class <- function(x, ...) {
  cat("my_class object\n")
  invisible(x)
}

Such methods do not need @return or @examples, and R CMD check does not flag them as undocumented (registered S3 methods are excluded from the undocumented-objects check).

  • The exemption holds only while the block contains nothing but @export (or @exportS3Method). Adding a title, @param, @rdname, @describeIn, or any other documentation makes roxygen2 generate an .Rd file, and the function must then satisfy all documentation rules including @return.
  • Do not confuse these with the S3 generic itself (e.g., my_print <- function(x, ...) UseMethod("my_print")). A newly exported generic is an ordinary exported function with documentation, so it still needs @return and @examples.
  • Methods sharing a page with their generic via @rdname or @describeIn follow the same rules as the generic: the combined .Rd needs the required tags once, not once per method.

Example:

# Missing @return - WILL BE REJECTED
#' Calculate sum
#' @export
my_sum <- function(x, y) {
  x + y
}

# Correct - includes @return
#' Calculate sum
#' @param x First number
#' @param y Second number
#' @return A numeric value
#' @export
my_sum <- function(x, y) {
  x + y
}

# For functions with no return value
#' Print message
#' @param msg Message to print
#' @return None, called for side effects
#' @export
print_msg <- function(msg) {
  cat(msg, "\n")
}

Examples for Exported Functions

If your exported function has a meaningful return value, it will almost definitely require an @examples section. Use the roxygen2 tag @examples.

  • Required even for functions marked @keywords internal
  • Exceptions exist for functions used purely for side effects (e.g., creating directories)
  • Examples must be executable
  • S3 methods with @export-only roxygen blocks have no .Rd file and need no examples (see above)

Un-exported Functions with Examples

If you write roxygen examples for un-exported functions, you must either:

  1. Call them with ::: notation: pkg:::my_fun()
  2. Use @noRd tag to suppress .Rd file creation

Using \dontrun{} Sparingly

\dontrun{} should only be used if the example really cannot be executed (e.g., missing additional software, API keys, etc.).

  • If showing an error, wrap the call in try() instead
  • Consider custom predicates (e.g., googlesheets4::sheets_has_token()) with if () blocks
  • Sometimes interactive() can be used as the condition
  • Lengthy examples (> 5 sec) can use \donttest{}

Never Comment Out Code in Examples

# BAD - Will be rejected
#' @examples
#' # my_function(x)  # Don't do this!

CRAN's guidance: "Examples/code lines in examples should never be commented out. Ideally find toy examples that can be regularly executed and checked."

Guarding Examples with Suggested Packages

Use @examplesIf for entire example sections requiring suggested packages:

#' @examplesIf rlang::is_installed("dplyr")
#' library(dplyr)
#' my_data %>% my_function()

For individual code blocks within examples:

#' @examples
#' if (rlang::is_installed("dplyr")) {
#'   library(dplyr)
#'   my_data %>% my_function()
#' }

DESCRIPTION Title Field

CRAN enforces strict Title requirements:

Use Title Case

Capitalize all words except articles like 'a', 'the'. Use tools::toTitleCase() to help format.

Avoid Redundancy

Common phrases that get flagged:

  • "A Toolkit for" → Remove
  • "Tools for" → Remove
  • "for R" → Remove

Examples:

# BAD
Title: A Toolkit for the Construction of Modeling Packages for R

# GOOD
Title: Construct Modeling Packages

# BAD
Title: Command Argument Parsing for R

# GOOD
Title: Command Argument Parsing

Quote Software/Package Names

Put all software and R package names in single quotes:

# GOOD
Title: Interface to 'Tiingo' Stock Price API

Length Limit

Keep titles under 65 characters.

DESCRIPTION Description Field

Never Start With Forbidden Phrases

CRAN will reject descriptions starting with:

  • "This package"
  • Package name
  • "Functions for"
# BAD
Description: This package provides functions for rendering slides.
Description: Functions for rendering slides to different formats.

# GOOD
Description: Render slides to different formats including HTML and PDF.

Expand to 3-4 Sentences

Single-sentence descriptions are insufficient. Provide a broader description of:

  • What the package does
  • Why it may be useful
  • Types of problems it helps solve
# BAD (too short)
Description: Render slides to different formats.

# GOOD
Description: Render slides to different formats including HTML and PDF.
    Supports custom themes and progressive disclosure patterns. Integrates
    with 'reveal.js' for interactive presentations. Designed for technical
    presentations and teaching materials.

Quote Software Names, Not Functions

# BAD
Description: Uses 'case_when()' to process data.

# GOOD
Description: Uses case_when() to process data with 'dplyr'.

Software, package, and API names get single quotes (including 'R'). Function names do not.

Expand All Acronyms

All acronyms must be fully expanded on first mention:

# BAD
Description: Implements X-SAMPA processing.

# GOOD
Description: Implements Extended Speech Assessment Methods Phonetic
    Alphabet (X-SAMPA) processing.

Publication Titles Only in Double Quotes

Only use double quotes for publication titles, not for phrases or emphasis:

# BAD
Description: Handles dates like "the first Monday of December".

# GOOD
Description: Handles dates like the first Monday of December.

All URLs Must Use HTTPS

CRAN requires https:// protocol for all URLs. HTTP links will be rejected.

# BAD
URL: http://paleobiodb.org/

# GOOD
URL: https://paleobiodb.org/

No Redirecting URLs

CRAN rejects URLs that redirect to other locations. Example rejection:

Found the following (possibly) invalid URLs:
URL: https://h3geo.org/docs/core-library/coordsystems#faceijk-coordinates
     (moved to https://h3geo.org/docs/core-library/coordsystems/)

Use urlchecker Package

# Find redirecting URLs
urlchecker::url_check()

# Automatically update to final destinations
urlchecker::url_update()

Ignore URLs That Will Exist After Publication

Some URLs that don't currently resolve will exist once the package is published on CRAN. These should NOT be changed:

  • CRAN badge URLs (e.g., https://cran.r-project.org/package=pkgname)
  • CRAN status badges (e.g., https://www.r-pkg.org/badges/version/pkgname)
  • CRAN check results (e.g., https://cranchecks.info/badges/pkgname)
  • Package documentation URLs on r-universe or pkgdown sites that deploy after release

When urlchecker::url_check() flags these URLs, leave them as-is. They are aspirational URLs that will work once the package is on CRAN.

Check for Invalid File URIs

Relative links in README must exist after package build. Common issue:

Found the following (possibly) invalid file URI:
     URI: CODE_OF_CONDUCT.md
       From: README.md

This occurs when files are in .Rbuildignore. Solutions:

  1. Remove file from .Rbuildignore
  2. Use usethis::use_code_of_conduct() which generates sections without relative links

Administrative Requirements

Copyright Holder Role

Always add [cph] role to Authors field, even if you're the only author:

# Required
Authors@R: person("John", "Doe", role = c("aut", "cre", "cph"))

Posit-Supported Packages

For packages in Posit-related GitHub organizations (posit-dev, rstudio, r-lib, tidyverse, tidymodels) or maintained by someone with a @posit.co email address, include Posit Software, PBC as copyright holder and funder:

Authors@R: c(
  person("Jane", "Doe", role = c("aut", "cre"),
         email = "[email protected]"),
  person("Posit Software, PBC", role = c("cph", "fnd"),
         comment = c(ROR = "03wc8by49"))
)

LICENSE Year

Update LICENSE year to current submission year:

# If LICENSE shows 2024 but submitting in 2026
# Update: 2024 → 2026

Method References

CRAN may ask:

If there are references describing the methods in your package, please add these in the description field...

If there are no references, reply to the email explaining this. Consider adding a preemptive note in cran-comments.md:

## Method References

There are no published references describing the methods in this package.
The package implements original functionality for [brief description].

Key Files to Review

Work through these files systematically:

  • DESCRIPTION: Title, Description, Authors@R, URLs, License year
  • R/*.R: Function documentation (@return, @examples, @examplesIf, @noRd)
  • README.Rmd (if exists): Edit this file (NOT README.md), then run devtools::build_readme()
  • README.md: Review for install instructions, relative links, URLs. Only edit directly if no README.Rmd exists
  • cran-comments.md: Preemptive notes for reviewers
  • NEWS.md: Version notes for this release
  • .Rbuildignore: Files referenced in README

Common Fix Patterns

DESCRIPTION Title:

# Before
Title: A Toolkit for the Construction of Modeling Packages for R

# After
Title: Construct Modeling Packages

DESCRIPTION Description:

# Before
Description: This package provides functions for rendering slides.

# After
Description: Render slides to different formats including HTML and PDF.
    Supports custom themes and progressive disclosure. Integrates with
    'reveal.js' for interactive presentations.

Function Documentation:

# Before - Missing @return
#' Calculate total
#' @param x Values
#' @export
calc_total <- function(x) sum(x)

# After - Complete documentation
#' Calculate total
#' @param x Numeric values to sum
#' @return A numeric value representing the sum
#' @examples
#' calc_total(1:10)
#' @export
calc_total <- function(x) sum(x)

Useful Tools

  • tools::toTitleCase() - Format titles with proper capitalization
  • urlchecker::url_check() - Find problematic URLs
  • urlchecker::url_update() - Fix redirecting URLs
  • usethis::use_news_md() - Create NEWS.md
  • usethis::use_cran_comments() - Create cran-comments.md
  • devtools::build_readme() - Re-render README.md from README.Rmd
  • usethis::use_code_of_conduct() - Add CoC without relative links
  • usethis::use_build_ignore() - Ignore files in R package build
  • usethis::use_package() - Add a package dependency to DESCRIPTION
  • usethis::use_tidy_description() - Tidy up DESCRIPTION formatting

Final Verification Checklist

Use this checklist to ensure nothing is missed before submission:

Files and Structure

  • NEWS.md exists and documents changes for this version
  • cran-comments.md exists with submission notes
  • If README.Rmd exists, it was edited (not README.md) and devtools::build_readme() was run
  • README includes valid install instructions (install.packages("pkgname"))
  • README has no relative links (all links are full URLs or removed)

DESCRIPTION File

  • Title: uses title case
  • Title: has no redundant phrases ("A Toolkit for", "Tools for", "for R")
  • Title: quotes all software/package names in single quotes
  • Title: is under 65 characters
  • Description: does NOT start with "This package", package name, or "Functions for"
  • Description: is 3-4 sentences explaining purpose and utility
  • Description: quotes software/package/API names (including 'R') but NOT function names
  • Description: expands all acronyms on first mention
  • Description: uses double quotes only for publication titles
  • Authors@R: includes copyright holder with [cph] role
  • For Posit packages: Includes person("Posit Software, PBC", role = c("cph", "fnd"), comment = c(ROR = "03wc8by49"))
  • LICENSE year matches current submission year

Function Documentation

  • All exported functions with .Rd files have @return documentation
  • All documented exported functions with meaningful returns have @examples
  • S3 methods with @export-only roxygen blocks are correctly exempt (no .Rd file, no @return/@examples needed)
  • S3 generics added by the package have full documentation (@return, @examples) since they are ordinary exported functions
  • No example sections use commented-out code
  • Examples avoid \dontrun{} unless truly necessary
  • Examples requiring suggested packages use @examplesIf or if guards
  • Un-exported functions with examples use ::: notation or @noRd
  • urlchecker::url_check() was run
  • All URLs use https protocol (no http links)
  • No redirecting URLs (except aspirational CRAN badge URLs)
  • Aspirational URLs (CRAN badges, etc.) are left as-is
  • No relative links in README that reference .Rbuildignore files
  • If concerns about method references, added preemptive note to cran-comments.md
  • Reviewed bundled file licensing if including third-party files

相似的 Skill

brand-guidelines
anthropics/skills180k

brand-guidelines

Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.

AI 与智能体

internal-comms
anthropics/skills180k

internal-comms

A set of resources to help me write all kinds of internal communications, using the formats that my company likes to use. Claude should use this skill whenever asked to write some sort of internal communications (status reports, leadership updates, 3P updates, company newsletters, FAQs, incident reports, project updates, etc.).

AI 与智能体

template-skill
anthropics/skills180k

template-skill

Replace with description of the skill and when Claude should use it.

AI 与智能体

mcp-builder
anthropics/skills180k

mcp-builder

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

AI 与智能体

algorithmic-art
anthropics/skills180k

algorithmic-art

Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.

AI 与智能体

academy-guide
anthropics/skills180k

academy-guide

Stop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.

AI 与智能体