forrtproject · richarddushime · Jul 5, 2026
@@ -3,7 +3,21 @@
     "allow": [
       "Bash(wc:*)",
       "Bash(git:*)",
-      "Bash(gh pr:*)"
+      "Bash(gh pr:*)",
+      "Bash(python3 -c \"import yaml; print\\(yaml.__version__\\)\")",
+      "Bash(pip3 show *)",
+      "Bash(pip3 install *)",
+      "Bash(python3 -c \"from ruamel.yaml import YAML; print\\('ok'\\)\")",
+      "Bash(mkdir -p /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest/data /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest/scripts)",
+      "Bash(cp /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/publications_test.yaml /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest/data/publications.yaml)",
+      "Bash(cp /Users/rdm/oss/forrtproject.github.io/scripts/build_publications.py /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest/scripts/)",
+      "Bash(python3 scripts/build_publications.py)",
+      "Bash(mkdir -p /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest2/data /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest2/scripts)",
+      "Bash(cp /Users/rdm/oss/forrtproject.github.io/scripts/build_publications.py /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest2/scripts/)",
+      "Bash(cat)",
+      "Bash(mkdir -p /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest3/data /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest3/scripts)",
+      "Bash(cp /Users/rdm/oss/forrtproject.github.io/scripts/build_publications.py /private/tmp/claude-501/-Users-rdm-oss-forrtproject-github-io/d40a3858-03b0-4abe-8e07-eb52ecb3b04b/scratchpad/pubtest3/scripts/)",
+      "Bash(python3 scripts/build_publications.py --dry-run)"
     ]
   }
 }
@@ -72,6 +72,21 @@ Processes:
 
 Triggers `deploy.yaml` after successful processing via `repository_dispatch`.
 
+### [update_publications.yml](workflows/update_publications.yml)
+**Hydrates DOI-stub entries in `data/publications.yaml`**
+
+| Trigger | Schedule |
+|---------|----------|
+| Push to `main` touching `data/publications.yaml` | On push |
+| Weekly | Monday 03:30 UTC |
+| Manual dispatch | Manual (`force` option to bypass the DOI cache) |
+
+Runs `scripts/build_publications.py`, which fetches title/authors/journal/year/
+abstract/citation for any entry with a top-level `doi:` key from Crossref/DataCite
+(via doi.org content negotiation). Never commits to `main` directly — opens a PR
+for review when metadata changes, same pattern as the glossary-update PR in
+`data-processing.yml`. See `content/publications/README` for the entry format.
+
 ## Quality Checks
 
 ### [spell-check.yaml](workflows/spell-check.yaml)

@@ -0,0 +1,90 @@
+name: Update publications
+
+# =====================================================================
+# Hydrates DOI-stub entries in data/publications.yaml (see
+# scripts/build_publications.py) by fetching title/authors/journal/year/
+# abstract/citation from the DOI's registration agency (Crossref/DataCite).
+#
+# Never commits to main directly: if hydration produces a diff, it's pushed
+# to a new branch and opened as a PR for human review before it goes live.
+#
+# Triggers:
+#   - push to main touching data/publications.yaml (hydrate a newly added
+#     DOI stub right away)
+#   - weekly schedule (catch metadata drift: preprint -> published,
+#     volume/page assignment, etc.)
+#   - workflow_dispatch (manual run, optionally ignoring the DOI cache)
+# =====================================================================
+
+on:
+  push:
+    branches: [main]
+    paths:
+      - 'data/publications.yaml'
+  schedule:
+    - cron: '30 3 * * 1'   # weekly, Monday 03:30 UTC
+  workflow_dispatch:
+    inputs:
+      force:
+        description: 'Ignore the DOI lookup cache and refetch everything'
+        required: false
+        type: boolean
+        default: false
+
+concurrency:
+  group: update-publications
+  cancel-in-progress: false
+
+jobs:
+  hydrate:
+    name: Hydrate publication DOIs
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v6
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Install dependencies
+        run: pip install ruamel.yaml
+
+      - name: Restore DOI metadata cache
+        uses: actions/cache@v4
+        with:
+          path: scripts/publications_doi_cache.json
+          key: publications-doi-cache-${{ github.run_id }}
+          restore-keys: |
+            publications-doi-cache-
+
+      - name: Run hydration script
+        run: |
+          python3 scripts/build_publications.py ${{ inputs.force && '--force' || '' }}
+
+      - name: Create PR with hydrated metadata
+        env:
+          GITHUB_TOKEN: ${{ secrets.FORRT_PAT }}
+        run: |
+          if [ -z "$(git status --porcelain -- data/publications.yaml)" ]; then
+            echo "ℹ️ No publication metadata changes detected"
+            exit 0
+          fi
+
+          BRANCH_NAME="publications-update-$(date +%Y%m%d-%H%M%S)"
+          git checkout -b "$BRANCH_NAME"
+
+          git add data/publications.yaml
+          git commit -m "Update hydrated publication metadata - $(date -u +'%Y-%m-%d %H:%M:%S UTC')"
+
+          git push -u origin "$BRANCH_NAME"
+
+          gh pr create \
+            --title "📚 Publications metadata update - $(date '+%Y-%m-%d')" \
+            --body "Automated DOI metadata hydration generated on $(date -u +'%Y-%m-%d %H:%M:%S UTC') by scripts/build_publications.py. Please review the hydrated title/authors/abstract/citation fields before merging." \
+            --base main \
+            --head "$BRANCH_NAME"
+          echo "✅ PR created for publications update"
@@ -277,3 +277,6 @@ bibtex_to_apa/node_modules/
 
 # DOI lookup cache for cluster parsing script
 scripts/doi_cache.json
+
+# DOI metadata cache for publications hydration script
+scripts/publications_doi_cache.json
@@ -2,9 +2,60 @@
 
 This folder contains PDF files and other publication-related assets. The actual publication data is stored in `/data/publications.yaml` (NOT in this folder).
 
-## Adding a New Publication
+## Adding a New Publication (DOI, recommended)
 
-Publications are managed in the **`/data/publications.yaml`** file. This file contains all publication metadata that is displayed on the website.
+If the publication has a DOI, you only need to add a minimal stub — a script
+fetches the rest. Add an entry with a top-level `doi:` key:
+
+```yaml
+- doi: "10.1016/j.socec.2025.102502"
+  type: "journal"          # journal, preprint, media, policy, affiliated, wip
+  focus_area: "Meta-Research & Scientific Reform"
+```
+
+Commit that stub to `data/publications.yaml` and push (or open a PR). The
+[update_publications.yml](/.github/workflows/update_publications.yml) workflow
+runs `scripts/build_publications.py`, which resolves the DOI via Crossref/DataCite
+and fills in `title`, `authors`, `journal_name`, `year`, `status`, `abstract`,
+`citation`, `links.doi`, and `altmetric_doi`. It opens a PR with the hydrated
+result for review — nothing is pushed straight to `main`.
+
+**On every run, all of the auto-filled fields above are recomputed from the DOI
+and will overwrite hand edits.** To pin a field permanently (a custom abstract
+blurb, an extra `postprint`/`pdf`/`project` link, a manually-chosen `status`),
+add it under `overrides:` instead of editing the auto field directly:
+
+```yaml
+- doi: "10.1016/j.socec.2025.102502"
+  type: "journal"
+  focus_area: "Meta-Research & Scientific Reform"
+  overrides:
+    abstract: "Custom short blurb instead of the paper's full abstract"
+    status: "In Press"
+    links:
+      postprint: "https://osf.io/preprints/..."
+      pdf: "./pdf/local-copy.pdf"
+```
+
+`type` and `focus_area` are never touched by the script — always set those by
+hand. If a DOI fails to resolve (bad DOI, API outage), the script leaves that
+entry's existing fields untouched and logs a warning rather than blanking them.
+
+Citations are built as
+`{authors} ({year}). <strong>{title}.</strong> <em>{journal_name}</em>, {volume}({issue}), {page}. https://doi.org/{doi}`
+— good for typical journal articles and preprints, but book chapters and other
+unusual venues often need a hand-written `overrides.citation`. Check the PR
+diff before merging.
+
+**Note:** Existing publications that predate this pipeline are still fully
+hand-written (no top-level `doi:` key) and are left alone by the script —
+migrating them to the DOI-stub format is a separate, deliberate task, not
+something that happens automatically.
+
+## Adding a New Publication (manual, no DOI)
+
+For publications without a DOI (op-eds, policy briefs, work-in-progress, etc.),
+write the full entry by hand as before.
 
 ### Steps to Add a Publication
 

@@ -1,6 +1,23 @@
 # Below is a template for adding a new publication to the publications.yaml file. in data/publications.yaml
 # Remove or add the fields as needed following the layout of the existing entries.
 
+# --- Preferred: DOI stub (auto-hydrated by scripts/build_publications.py) ---
+# Only `doi`, `type`, and `focus_area` are hand-written; the rest is fetched
+# from Crossref/DataCite and refreshed on every run. Pin a field permanently by
+# adding it under `overrides:` instead. See content/publications/README.
+
+- doi: "10.xxxx/xxxxx"
+  type: "journal"   # journal, preprint, media, policy, affiliated, wip
+  focus_area: "Meta-Research & Scientific Reform"
+  # overrides:
+  #   abstract: "Custom short blurb instead of the paper's full abstract"
+  #   status: "In Press"
+  #   links:
+  #     postprint: "https://osf.io/preprints/..."
+  #     pdf: "./pdf/local-copy.pdf"
+
+# --- Fallback: full manual entry (no DOI, e.g. op-eds, policy briefs, WIP) ---
+
 - title: ""
   type: ""
   year: ""

@@ -8,3 +8,4 @@ gspread
 python-dotenv
 matplotlib
 pypinyin
+ruamel.yaml