Remove the five scripts that never ran, and guard the cause (#410, #523)

.gitignore

@@ -90,7 +90,7 @@ config.local.mk
 *.log
 
 # Curation / audit pipeline outputs — these are regeneratable artifacts
-# produced by scripts/curate_evidence_with_pdfs.py and the
+# produced by scripts/curate_evidence_with_pdfs.py (removed in #410) and the
 # `audit-network-report` justfile recipe; downstream scripts read them
 # within a workflow run, but they should not live in version control.
 evidence_curation_report.txt

docs/AUTOMATION_TOOLS.md

@@ -198,10 +198,11 @@ uv run python scripts/intelligent_snippet_fixer.py --file FILENAME.yaml
 # 2. Review suggestions carefully, apply best ones
 
 # 3. Validate
-# ⚠️ NOT FUNCTIONAL — this script imports communitymech.literature_enhanced,
-# which has never existed, so it fails at import. Tracked in #410.
-# (The command below is also stale: this repo uses uv, not poetry.)
-uv run python scripts/curate_evidence_with_pdfs.py --file FILENAME.yaml
+# ⚠️ REMOVED in #410. curate_evidence_with_pdfs.py never worked — it imported
+# communitymech.literature_enhanced, a module absent from every commit — and the
+# file itself is now gone, so this is what to run instead:
+uv run python scripts/cache_fulltext.py PMID:12345   # only if snippets need full text
+just validate-references kb/communities/FILENAME.yaml
 
 # 4. Schema check
 just validate kb/communities/FILENAME.yaml

docs/CURATION_PROGRESS_REPORT.md

@@ -259,7 +259,7 @@ poetry run python scripts/batch_snippet_fixer.py --phase 1 --auto-approve
 
 1. **Validation**:
    ```bash
-   # NOT FUNCTIONAL (#410): scripts/curate_evidence_with_pdfs.py --quick
+   # REMOVED in #410 (use cache_fulltext.py + just validate-references): scripts/curate_evidence_with_pdfs.py --quick
    just validate-all
    ```
 
@@ -313,10 +313,10 @@ poetry run python scripts/intelligent_snippet_fixer.py --file FILENAME.yaml --au
 poetry run python scripts/batch_snippet_fixer.py --phase 1 --auto-approve
 
 # Validate file
-# NOT FUNCTIONAL (#410): scripts/curate_evidence_with_pdfs.py --file FILENAME.yaml
+# REMOVED in #410 (use cache_fulltext.py + just validate-references): scripts/curate_evidence_with_pdfs.py --file FILENAME.yaml
 
 # Validate all files (quick)
-# NOT FUNCTIONAL (#410): scripts/curate_evidence_with_pdfs.py --quick
+# REMOVED in #410 (use cache_fulltext.py + just validate-references): scripts/curate_evidence_with_pdfs.py --quick
 ```
 
 ---

docs/pdf_fetching_capability.md

@@ -6,17 +6,35 @@ The CommunityMech literature system integrates a 6-tier cascading PDF discovery
 
 ## Integration Status
 
-⚠️ **NOT FUNCTIONAL** — the scripts described below cannot run.
+⚠️ **REMOVED** — the scripts described below no longer exist, and never worked.
 
 All of them begin `from communitymech.literature_enhanced import
 EnhancedLiteratureFetcher`, and that module has never existed in any commit: it
 was absent from the repo's first commit (`7c658e6`), which is also where these
 scripts were added. They fail at import, so even `--help` does not work.
 
+They were deleted in #410. Porting them was not an import swap: their CLI flags
+advertise a 6-tier PDF cascade with "fallback mirrors", and `LiteratureFetcher`
+has no PDF surface at all, so a port meant deciding whether to *build* that
+capability. The answer was no — retrieving publisher PDFs through mirrors is not
+something to add speculatively, and the real need behind it is open-access full
+text, which `scripts/cache_fulltext.py` already serves. The #183 sweep used it to
+cache full text for 64 of 125 references.
+
+**Use instead:**
+
+| Removed script | What to use |
+|---|---|
+| `test_pdf_fetching.py` | nothing — it tested the cascade that was never committed |
+| `curate_evidence_with_pdfs.py` | `scripts/cache_fulltext.py`, then `just validate-references` |
+| `quick_literature_review.py` | `just validate-references FILE` |
+| `review_literature.py` | `just validate-references FILE` |
+| `extract_evidence_snippets.py` | no direct replacement; snippets are curated by hand against the cache |
+
 This page previously read "✅ VALIDATED — Successfully integrated and tested",
 which was never true of the committed code.
 
-**Everything below this line describes software that is not in this repository**,
+**Everything below this line describes software that was never in this repository**,
 including the success rates, the per-tier table and the "5/5 DOIs" test results.
 Those numbers cannot have come from committed code and are retained only as a
 record of what was claimed. Do not cite them.