diff --git a/aborist/sources/crawler/bridge.py b/aborist/sources/crawler/bridge.py
index 969a079..fbb37f2 100644
--- a/aborist/sources/crawler/bridge.py
+++ b/aborist/sources/crawler/bridge.py
@@ -72,7 +72,10 @@ async def _crawl_seed_async(
         seen.add(url)
 
         try:
-            html, links = await fetcher.fetch_webpage(url)
+            # extract_links=True is REQUIRED for BFS — the verbatim
+            # AsyncWebFetcher defaults it to False (caller opts in for the
+            # link-graph use case). Without it BFS terminates at the seed.
+            html, links = await fetcher.fetch_webpage(url, extract_links=True)
         except Exception:
             continue
         if html is None:
diff --git a/tests/crawler/test_bridge.py b/tests/crawler/test_bridge.py
index 831cb25..d81b077 100644
--- a/tests/crawler/test_bridge.py
+++ b/tests/crawler/test_bridge.py
@@ -64,7 +64,7 @@ class _StubFetcher:
     def __init__(self, page_links: dict[str, list[str]]):
         self.page_links = page_links
 
-    async def fetch_webpage(self, url: str):
+    async def fetch_webpage(self, url: str, extract_links: bool = False, **kw):
         return "<html></html>", self.page_links.get(url, [])
 
 
@@ -137,6 +137,36 @@ def test_crawl_seed_dedups():
     assert len(out) == len(set(out))
 
 
+def test_crawl_seed_passes_extract_links_true():
+    """Regression: AsyncWebFetcher.fetch_webpage defaults extract_links=False,
+    so without the explicit kwarg BFS terminates after the seed (only one
+    URL ever lands in `discovered`). Pin the kwarg by recording the calls
+    on a stub fetcher."""
+
+    class _RecordingFetcher:
+        def __init__(self):
+            self.calls: list[dict] = []
+
+        async def fetch_webpage(self, url, extract_links=False, **kw):
+            self.calls.append({"url": url, "extract_links": extract_links})
+            return "<html></html>", []
+
+    import asyncio
+    from aborist.sources.crawler import bridge
+
+    fetcher = _RecordingFetcher()
+    asyncio.run(
+        bridge._crawl_seed_async(
+            "https://x.com/", max_depth=2, max_pages=5, fetcher=fetcher
+        )
+    )
+    assert fetcher.calls, "fetcher was never called"
+    for call in fetcher.calls:
+        assert call["extract_links"] is True, (
+            f"BFS must request link extraction; got {call!r}"
+        )
+
+
 # ---------------------------------------------------------------------------
 # ingest_crawled: stores ETag + Last-Modified per document_root
 # ---------------------------------------------------------------------------