feat(llm): corpus indexer skips items tagged llm:skip (refs #688)

src/llm/source.py

@@ -495,13 +495,17 @@ def _attachment_sections(
 def iter_corpus_refs(
     store: Store, *, tag: str = "", zotero: "ZoteroPdfIndex | None" = None
 ) -> Iterator[DocRef]:
-    """One DocRef per non-comment item. Fingerprint = updated_at + the
+    """One DocRef per non-comment, non-skipped item. Fingerprint =
-    bib attachment file stats; the Zotero fallback is only consulted by
+    updated_at + the bib attachment file stats; the Zotero fallback is
-    ``load()``."""
+    only consulted by ``load()``.
+
+    Items tagged ``llm:skip`` are excluded — the escape hatch for
+    material that must never enter the indexer (e.g. copyrighted CPT
+    editions imported for citation only, not for RAG text)."""
     sql = (
         "SELECT i.key, COALESCE(i.updated_at,'') AS updated_at FROM items i "
         "WHERE i.id NOT IN (SELECT item_id FROM item_tags WHERE tag_id IN "
-        "(SELECT id FROM tags WHERE name = 'doctype:comment'))"
+        "(SELECT id FROM tags WHERE name IN ('doctype:comment', 'llm:skip')))"
         + (
             " AND i.id IN (SELECT item_id FROM item_tags WHERE tag_id IN "
             "(SELECT id FROM tags WHERE name = ?))"

tests/llm/test_source.py

@@ -133,6 +133,13 @@ class TestCorpusDocs:
     def test_comments_excluded_from_corpus(self, store):
         assert all(d.metadata["doctype"] != "comment" for d in iter_corpus_docs(store))
 
+    def test_llm_skip_tag_excludes_item_from_corpus(self, store):
+        key = store.create(
+            Item(item_type="source", title="CPT edition", abstract="Descriptor text.")
+        )
+        store.add_tag(key, "llm:skip")
+        assert key not in {d.key for d in iter_corpus_docs(store)}
+
     def test_attachment_text_sectioned_and_listed_in_files(self, store, tmp_path):
         key = store.create(Item(item_type="rule", title="Rule with attachment"))
         store.add_tag(key, "year:2021")