diff --git a/cauldron/db.py b/cauldron/db.py
index eec29fe..8621eeb 100644
--- a/cauldron/db.py
+++ b/cauldron/db.py
@@ -2251,7 +2251,18 @@ class DB:
     ) -> int | None:
         """INSERT a freshly-scraped recipe in 'raw' state. Returns the new
         row id, or None if the source_url was already present (UNIQUE
-        violation = duplicate scrape, treat as skip)."""
+        violation = duplicate scrape, treat as skip).
+
+        Normalizes source_url by stripping trailing slashes so that
+        `.../recipes/falafel` and `.../recipes/falafel/` map to the same
+        UNIQUE key. 2026-05-02: caught when manual `/discover` paste
+        included trailing slash but listing-page extractor stripped it,
+        producing 1:1 duplicates."""
+        # URL canonicalization — single rstrip is safe for recipe paths
+        # (they always have a non-slash terminal segment; `https://host/`
+        # alone wouldn't be a valid recipe URL anyway).
+        if source_url.endswith("/"):
+            source_url = source_url.rstrip("/")
         with self.conn() as c, c.cursor() as cur:
             cur.execute(
                 """INSERT IGNORE INTO cauldron_discovered_recipes