Add the "reprexes" skill (.claude/skills/reprexes)

[d-morrison]

Summary

Adds a project-level Claude Code skill at .claude/skills/reprexes/SKILL.md.

reprexes captures a debugging workflow: when a technical problem is hard to pin down, isolate it into a minimal reproducible example ("reprex") and iterate candidate fixes on that fast, clean loop instead of inside the full application — then port the fix back. Includes the R/Quarto reprex package workflow (reprex::reprex(), venue / session_info / std_out_err).

References: https://r4ds.hadley.nz/workflow-help.html#making-a-reprex, https://reprex.tidyverse.org/.

Vendoring it here makes it available to collaborators and the @claude CI bot in this repo (previously it lived only in a personal user-level config).

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR adds a repository-local Claude Code skill (reprexes) documenting a workflow for isolating bugs into minimal reproducible examples (reprexes), with specific guidance for R and Quarto debugging.

Changes:

• Added a new .claude skill definition and documentation for creating and iterating on reprexes.
• Included R/Quarto-specific recommendations, including use of the {reprex} package and targeted quarto render workflows.

[github-actions]

PR Preview Action v1.8.1-2-g6ad689f
🚀 View preview at https://UCD-SERG.github.io/lab-manual/pr-preview/pr-282/
Built to branch gh-pages at 2026-06-17 17:26 UTC. Preview will be ready when the GitHub Pages deployment is complete.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 3 comments.

[d-morrison]

@claude review

[claude]

Claude finished @d-morrison's task in 1m 18s —— View job

---

Code Review

• Gather context
• Read SKILL.md file
• Review for correctness, clarity, and best practices
• Post review findings

---

Overall: This is a well-structured, accurate, and practical skill document. The workflow is sound, the R-specific details are correct, and the guidance is actionable. Two issues from the existing Copilot review comments remain unaddressed and are worth fixing before merge.

---

Issues to fix

1. Namespace inconsistency — reprex(input = ...) vs. reprex::reprex() (line 99)

The file introduces the function as reprex::reprex() (line 98) but then immediately drops the namespace qualifier:

reprex(input = "/tmp/reprex.R")

This will fail unless the package is explicitly attached with library(reprex) — which isn't shown. For consistency and copy-paste reliability, this should be:

reprex::reprex(input = "/tmp/reprex.R")

Fix this →

2. Rscript /tmp/reprex.R still loads user profiles (line 68)

Rscript /tmp/reprex.R

Rscript without flags still sources the user's .Rprofile and .Renviron, which can silently affect behavior (e.g., options, library paths, env vars). Since this step is explicitly framed as "run in a clean session," using:

Rscript --vanilla /tmp/reprex.R

would be more consistent with that goal. (--vanilla is equivalent to --no-save --no-restore --no-site-file --no-environ --no-init-file.)

Fix this →

---

Already addressed

• The earlier copilot note about tidyverse_update() is fully resolved — the current text correctly positions it as something to run outside the reprex to rule out stale packages, while explicitly noting it doesn't record versions.
• The "~80% of the time" unsourced statistic is no longer in the file.

---

Minor observations (no action required)

• The frontmatter (name, description, user-invocable, allowed-tools) is correctly structured.
• The "Before declaring the reprex good" checklist and the closing "Don't" section are both useful anchors.
• The reprex_clean() / reprex_rescue() / reprex_invert() companion helpers section is a nice addition not commonly documented in concise reprex guides.
• The cross-platform /tmp vs. tempdir() / tempfile() callouts are appropriately handled throughout.

[d-morrison]

Claude Code CLI (local session) is working on this — paws off until I'm done.

[d-morrison]

@claude review

[d-morrison]

Addressed both remaining findings from the latest review:

• Namespace consistency (line ~99): reprex(input = ...) → reprex::reprex(input = ...) so it works without an explicit library(reprex). (tidyverse::tidyverse_update() later in the file is already namespaced.)
• Clean-session flag (line ~68): Rscript /tmp/reprex.R → Rscript --vanilla /tmp/reprex.R, with a note that --vanilla skips .Rprofile/.Renviron, matching the stated clean-session goal.

The earlier inline notes (the ~80% statistic and the tidyverse_update() version-recording caveat) were already resolved in the current file per the review.

@claude review

[claude]

Claude finished @d-morrison's task in 1m 17s —— View job

---

Code Review

• Read SKILL.md
• Review for correctness, clarity, and best practices
• Post findings

---

Overall: The file is in good shape. Both issues flagged in the prior reviews (Copilot and previous Claude pass) have been resolved. No blocking issues remain.

---

Previously reported issues — now fixed

✅ reprex::reprex(input = ...) namespace — the call on what was line 99 now reads reprex::reprex(input = "/tmp/reprex.R"), consistent with the namespaced form introduced just above it.

✅ Rscript --vanilla — the clean-session step now correctly uses Rscript --vanilla /tmp/reprex.R, which suppresses .Rprofile / .Renviron loading as described.

✅ tidyverse::tidyverse_update() — properly namespaced, and the surrounding prose correctly notes it updates packages rather than recording versions, so it belongs outside the reprex itself.

✅ Unsourced statistic — the "~80% of the time" claim is gone.

---

Minor observations (no action required)

• The callr attribution for reprex 2.0's clean-session execution is accurate.
• session_info = TRUE correctly names both sessionInfo() and sessioninfo::session_info() — reprex uses whichever is available.
• std_out_err = TRUE capturing C-level / subprocess output is a useful edge-case note not commonly documented.
• reprex_clean() / reprex_rescue() / reprex_invert() are all real exported functions in reprex 2.x — the descriptions are correct.
• The dput() tactic for embedding minimal slices of real data inline is solid advice.
• The tempfile(fileext = ".R") / tempdir() cross-platform callouts appear consistently throughout — good.
• The closing "Don't" list is a useful anchor.

---

This is ready to merge from a content-accuracy standpoint.

[claude]

Claude finished @d-morrison's task in 1m 7s —— View job

---

Code Review

• Gather context
• Read current SKILL.md
• Verify claimed fixes
• Review for any remaining issues

---

Overall: Both claimed fixes are confirmed and correctly applied. The document is accurate, actionable, and ready to merge. No blocking issues remain.

---

Fixes verified

1. Namespace consistency — confirmed

Both calls are now correctly namespaced:

reprex::reprex()  # line ~98, clipboard variant
reprex::reprex(input = "/tmp/reprex.R")  # line ~99, file variant

No bare reprex(...) calls remain.

2. Clean-session flag — confirmed

Step 3 now reads:

Rscript --vanilla /tmp/reprex.R

And correctly explains: "--vanilla skips .Rprofile/.Renviron so no stale globalenv() state or profile settings mask or fake the bug." The rationale matches the stated goal.

---

Minor observations (no action required)

• The companion helpers reprex_clean(), reprex_rescue(), reprex_invert() are not namespaced, but they appear in descriptive prose rather than standalone copy-paste code, so the risk of confusion is much lower than the reprex::reprex(input = ...) case that was fixed.
• sessionInfo() is intentionally un-namespaced — it's a base R / utils function, so no qualifier is needed.
• The callr-since-2.0 note is accurate. The venue = table, std_out_err, wd =, and session_info = argument docs are all correct.
• The tidyverse::tidyverse_update() positioning (outside the reprex, to rule out stale packages, with an explicit note that it doesn't record versions) is exactly right.

---

This PR looks good to merge.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 3 comments.

[d-morrison]

Claude Code CLI (local session) is working on this — paws off until I'm done.

[d-morrison]

Addressed the minor observation.

Namespaced the companion helpers: reprex_clean(), reprex_rescue(), and reprex_invert() are now reprex::reprex_clean(), etc. I also namespaced the two other bare reprex() references (the reprex::reprex_invert() description and the "Validation bonus" line) so every reprex-package function in the doc is qualified consistently. They're all code-formatted (backticked) function references, so the reprex:: prefix reads naturally rather than oddly. sessionInfo() is left bare since it's base R, and tidyverse::tidyverse_update() was already qualified.

@claude review

[claude]

Claude finished @d-morrison's task in 2m 43s —— View job

---

Code Review

• Gather context
• Read current SKILL.md
• Verify claimed namespace fixes for companion helpers
• Check new Copilot comments (blank lines, "since reprex 2.0")
• Post findings

---

Overall: The claimed namespace fixes are confirmed. Three minor issues from the most recent Copilot pass (2026-06-03T22:22:05Z) remain unaddressed — two are trivial blank-line fixes, one is a factual accuracy question.

---

Claimed fix verified — ✅

All three companion helpers are now correctly namespaced:

reprex::reprex_clean()   (line 122)
reprex::reprex_rescue()  (line 123)
reprex::reprex_invert()  (line 125)

The two other bare reprex() references are also correctly qualified (reprex::reprex() at lines 99 and 117). No un-namespaced reprex-package calls remain.

---

Remaining issues (new Copilot comments, not yet addressed)

1. Missing blank line before the nested bullet list in step 2 (line ~62)

The sub-list after "Put in it, in order:" has no blank line separator:

   portable path). Put in it, in order:
   - the package loads (`library(...)`),   ← no blank line before this

Standard CommonMark / GitHub Markdown requires a blank line before a list inside a numbered list item for consistent rendering. Fix: add a blank line after Put in it, in order:.

Fix this →.%20Put%20in%20it%2C%20in%20order%3A%22%20add%20a%20blank%20line%20before%20the%20%22-%20the%20package%20loads%22%20nested%20bullet%20list%20to%20match%20the%20repo%E2%80%99s%20%22blank%20line%20before%20lists%22%20markdown%20formatting%20rule.&repo=UCD-SERG%2Flab-manual)

2. "since reprex 2.0" qualifier is likely inaccurate (line 97)

  session (via `callr` since reprex 2.0) and emits code **plus actual

The callr-backed subprocess execution was introduced in reprex 0.2.0 (2018), not reprex 2.0 (2021). reprex 2.0 was a major API rework, but it didn't introduce callr. The parenthetical is a factual claim that appears to be wrong, and it'll silently mislead readers over time. The safest fix is to drop the version qualifier: (via \callr`)`.

Fix this →

3. Missing blank line before "Useful arguments:" and before the venue = nested list (line ~103–105)

  output is destined for a PR comment or an upstream issue. Useful arguments:
  - `venue =` — output format:
    - `"gh"` — GitHub-flavored Markdown (default)   ← no blank line before nested list

Two spots need blank lines: before the - \venue =`list (or "Useful arguments:" should be its own paragraph), and before the nested list ofvenue` values. Without blank lines, some Markdown renderers may not recognize the nested structure.

Fix this →

---

Previously verified (no action needed)

All fixes from earlier review rounds are intact: reprex::reprex(input = ...), Rscript --vanilla, tidyverse::tidyverse_update() positioning, and the removed unsourced statistic.

---

The blank-line issues are cosmetic/render-safety; the callr version qualifier is the one substantive accuracy concern. Otherwise content is solid.

[d-morrison]

Claude Code CLI (local session) is working on this — paws off until I'm done.

[d-morrison]

Addressed the 3 remaining items from the latest Copilot pass (commit f296acd):

• Blank line before nested list in step 2 (line ~62) — added.
• Blank lines before "Useful arguments:" and the nested venue = list (line ~104) — added both.
• "since reprex 2.0" qualifier (line 97) — corrected. Verified against the reprex changelog: callr-backed clean-session execution dates to reprex 0.1.0 (2017), not 2.0. Dropped the version qualifier to (via callr) so the text stays accurate.

@claude review

[claude]

Claude finished @d-morrison's task in 53s —— View job

---

Code Review

• Gather context
• Read current SKILL.md
• Verify claimed fixes (blank lines, "since reprex 2.0" qualifier)
• Check for any remaining issues
• Post findings

---

Overall: All three claimed fixes are confirmed and correctly applied. The document is accurate, clean, and ready to merge. No blocking or minor issues remain.

---

Fixes verified — ✅ all three

1. Blank line before nested list in step 2

   portable path). Put in it, in order:

   - the package loads (`library(...)`),

Blank line is present. ✅

2. Blank lines before "Useful arguments:" list and nested venue = values

  output is destined for a PR comment or an upstream issue. Useful arguments:

  - `venue =` — output format:

    - `"gh"` — GitHub-flavored Markdown (default)

Both blank lines are present. ✅

3. "since reprex 2.0" qualifier dropped

The line now reads:

  session (via `callr`) and emits code **plus actual output**.

Version qualifier is gone. ✅

---

Full document check (no action needed)

• All reprex-package function calls are namespaced (reprex::reprex(), reprex::reprex_clean(), reprex::reprex_rescue(), reprex::reprex_invert()). ✅
• Rscript --vanilla /tmp/reprex.R is correct and matches the clean-session framing. ✅
• sessionInfo() correctly left bare (base R / utils). ✅
• tidyverse::tidyverse_update() correctly namespaced and positioned outside the reprex. ✅
• Cross-platform /tmp vs. tempfile(fileext = ".R") / tempdir() callouts are consistent throughout. ✅
• The reprex_clean() / reprex_rescue() / reprex_invert() descriptions are accurate. ✅
• The "Before declaring the reprex good" checklist and closing "Don't" section are both solid anchors. ✅

---

This PR is ready to merge.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated no new comments.

[d-morrison]

Done with my local session — unclaiming. (Re-triggered @claude review on the latest commit.)

[claude]

Claude finished @d-morrison's task in 1m 15s —— View job

---

Code Review

• Gather context
• Read current SKILL.md (latest commit f296acd)
• Verify all previously reported fixes are still in place
• Check for any new issues
• Post findings

---

Overall: All previously reported fixes are confirmed in place. The latest Copilot pass (2026-06-03T22:58:09Z) had no new comments. No blocking or minor issues remain — this PR is ready to merge.

---

All fixes verified — ✅

Namespace consistency
Every reprex-package function reference is qualified:

• reprex::reprex() (clipboard call) ✅
• reprex::reprex(input = "/tmp/reprex.R") (file call) ✅
• reprex::reprex_clean(), reprex::reprex_rescue(), reprex::reprex_invert() ✅
• sessionInfo() correctly left bare (base R / utils) ✅
• tidyverse::tidyverse_update() correctly namespaced and positioned outside the reprex ✅

Clean-session flag
Step 3 reads Rscript --vanilla /tmp/reprex.R with the correct rationale ("skips .Rprofile/.Renviron so no stale globalenv() state or profile settings mask or fake the bug") ✅

callr version qualifier dropped
The line now reads (via \callr`)` — the inaccurate "since reprex 2.0" qualifier is gone ✅

Blank lines before nested lists

• Blank line after "Put in it, in order:" (step 2) ✅
• Blank line before the venue = list and before its nested values ✅

Unsourced statistic — the "~80% of the time" claim is gone ✅

---

Full document check (no action needed)

• Frontmatter (name, description, user-invocable, allowed-tools) is correctly structured.
• Cross-platform /tmp vs. tempfile(fileext = ".R") / tempdir() callouts are consistent throughout.
• The dput() tactic for embedding minimal data slices inline is sound advice.
• The "Before declaring the reprex good" checklist and closing "Don't" section are both solid anchors.
• The reprex::reprex_invert() description ("the inverse of reprex::reprex()") is accurate.
• std_out_err = TRUE capturing C-level / subprocess output is a useful edge-case note.
• tidyverse::tidyverse_update() positioning (outside the reprex, to rule out stale packages, with an explicit note it doesn't record versions) is exactly right.
• Build-artifact confounders (_site/, _freeze/, .quarto/) callout is appropriate for a Quarto-heavy repo.

---

This PR is ready to merge.

[d-morrison]

Claude Code CLI (local session) is working on this — paws off until I'm done.