[LukasWallrich]

Applies the content-review decisions for the cluster pages: corrections to the Clusters v4.1 Google Doc, regenerated with two parser fixes, and cleaner publication titles in the featured-resources generator.

Clusters Doc and regenerated pages

The Doc was corrected first. Then parse_clusters_to_sheet.py --json-only and generate_cluster_pages.py were re-run.

• Names: clusters 5, 6 and 8 now use title case, matching the /resources/ filters ("FAIR Data and Materials", "Replication and Meta-Research", "Transparency and Reproducibility in Computation and Analysis").
• Descriptions:
  • WEIRD now expands "Industrialized".
  • The GOLD and DIAMOND open access glosses are corrected.
  • Rewritten: the equity definition, the LGBTQ+ scope, the communication-barriers list, the Study Design and qualitative-reporting sentences, the Netherlands Code sentence and the cluster 6 grammar.
  • Removed: the repeated NHST clause and the QRP fragment.
  • Rights retention gets an explanatory sentence.
  • Decolonizing Research Practices gets a description.
• Removed from the Doc: the "INSERT DESCRIPTION" placeholders, an editor's "To add:" note (now the Flynn, 2025 reference) and a Wikipedia note.
• References:
  • A duplicate Bartoš et al. (2024) is gone.
  • Mozersky et al. (2020) in IASSIST Quarterly now has its own DOI, 10.29173/iq952, checked against Crossref.
  • The URL-encoded DOI 10.1037%2Fqup0000076 is normalised.

Parser fixes (scripts/parse_clusters_to_sheet.py)

• Descriptions that cite a year: a description line that mentions a year in passing ("Clarke's (2022) piece") stays the description, so it is no longer turned into a reference. Lines that start like an author list, or that carry a DOI or URL, are still read as references. On the earlier Doc text, only the cluster 10 "Standards and guidelines for rigorous qualitative research" sub-cluster changes.
• Group authors: when the doi.org APA text starts with the year, the Doc's author is kept. This affects Open Science Collaboration (2012, 2015) and National Academies (2017).
• Tests are added for both fixes.

Publication titles (scripts/build_featured_data.py)

• Titles and APA references are now stored as plain text plus <i>/<scp> markup. HTML entities are decoded, including double-escaped <em>; <em> becomes <i>; and whitespace is collapsed.
• This also repairs citation buttons such as "Lakens & amp; Lieck (2022)".
• A separate PR into rebrand-doors renders the allow-listed tags.
• The change goes live with the next daily data-processing run.

AGENTS.md

AGENTS.md now points to --json-only. Run on its own, --export-json also clears and rewrites the sheet tabs, including the enriched Publications columns.

Checks: unit tests pass, hugo --gc --minify builds, and check_seo_html.py passes.

🤖 Generated with Claude Code

---

Summary by cubic

Applies the content-review decisions for the cluster pages and fixes two parser bugs that surfaced when regenerating them: clusters 5, 6, and 8 now use title case, confusing or broken descriptions and references were corrected, and the parser no longer mistakes prose that cites a year for a reference or drops group authors like Open Science Collaboration.

Parser fixes

• A description line mentioning a year in passing ("Clarke's (2022) piece") stays a description, provided it carries no "(year)." reference date; author-list, DOI, and URL lines are still treated as references.
• When the doi.org APA text starts with the year, the Doc's group author is kept (Open Science Collaboration, National Academies), including the cluster 8 2015 Science entry.
• The script's --help, AGENTS.md, and the site's data-error message now recommend --json-only, since plain --export-json also clears and rewrites the Clusters, Sub-Clusters, and Publications sheet tabs.

Content

• Corrected WEIRD's expansion, the GOLD and DIAMOND open access glosses, equity, LGBTQ+ scope, communication barriers, Study Design, and other cluster 6 descriptions.
• Removed a duplicate Bartoš et al. reference, the "INSERT DESCRIPTION" placeholders, and the Wikipedia and "To add:" notes; added a description for Decolonizing Research Practices.
• Publication titles and APA references now decode HTML entities and keep <i>/<scp> markup while leaving all other angle brackets escaped, preserving the full stop in "et al." and dropping it after group authors; this fixes broken citation buttons like "Lakens & Lieck (2022)".

^{Written for commit dbec563. Summary will update on new commits.}

[github-actions]

👍 All image files/references (if any) are in webp format, in line with our policy.

[github-actions]

✅ Spell Check Passed

No spelling issues found when checking 9 changed file(s)! 🎉

[LukasWallrich]

⚠️ Staging Deployment Status

This PR was attempted for staging deployment but had merge conflicts and was skipped.

Attempted at: 2026-11-13 12:30:41 UTC
Staging URL: https://staging.forrt.org

Please resolve conflicts with the base branch and the deployment will be retried automatically.

[cubic-dev-ai]

All reported issues were addressed across 9 files

<sub>Reply with feedback, questions, or to request a fix.

View guided diff | Turn on auto-fix | Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 6 files (changes from recent commits).

<sub>Reply with feedback, questions, or to request a fix.

View guided diff | Turn on auto-fix | Re-trigger cubic</sub>

[LukasWallrich]

Note

🤖 Claude Opus 5.5 responding on behalf of Lukas Wallrich

@richarddushime, could you review the code in this PR? The cluster data and page changes come from the Clusters Doc fixes Lukas approved, but these three code changes alter what the daily data run writes to forrt.org:

1. scripts/build_featured_data.py, clean_markup_text(). Titles and APA references in featured_resources.json / pub_cards.json now decode entities up to twice, turn <em> into <i>, and keep raw <i>/<scp> while re-encoding every other < and >. Is that the right contract for the data artifact? main's current templates escape titles, so <i> still shows as literal text there until the Doors templates (Render italics and small caps in publication titles on cluster pages #1083) arrive. Do you see any consumer on main that would output these fields unescaped?

2. scripts/parse_clusters_to_sheet.py. Two changes:

   • is_description_with_year(): a description line that cites a year in passing now stays the description.
   • keep_doc_author(): a group author such as "Open Science Collaboration." is prepended when the doi.org APA text starts with the year.

   Are the heuristics (AUTHOR_START_RE, REFERENCE_DATE_RE) narrow enough? The tests are in scripts/tests/test_parse_clusters.py.

3. AGENTS.md and the --help / data-error text now say to regenerate with --json-only, because --export-json on its own also clears and rewrites the Clusters, Sub-Clusters and Publications sheet tabs. Is that the workflow you want documented?