diff --git a/tests/test_html_list_rendering.py b/tests/test_html_list_rendering.py
new file mode 100644
index 0000000..dbc47e6
--- /dev/null
+++ b/tests/test_html_list_rendering.py
@@ -0,0 +1,85 @@
+"""Regression: the HTML reader must not collapse bold-lead-in lists into a paragraph.
+
+The book style writes a bold label immediately followed by a list, with no blank
+line between them:
+
+    **触发分离的条件**：
+    - 输入序列长度 > 某阈值
+
+Strict CommonMark (pandoc's default `-f markdown`) requires a blank line before a
+list, so it folds all three bullets into one <p>. build_html_reader.py therefore
+must pass `markdown+lists_without_preceding_blankline`. This test feeds that exact
+pattern through the *real* pandoc invocation the reader uses and asserts the list
+survives — the check that would have caught the 640 collapsed lists that shipped.
+"""
+
+import re
+import shutil
+import subprocess
+import unittest
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[1]
+READER = ROOT / "tools" / "build_html_reader.py"
+
+# The pattern the book uses everywhere: bold lead-in, then a list, no blank line.
+SAMPLE = "**触发分离的条件**：\n- 输入序列长度 > 某阈值\n- 输入序列未命中前缀缓存\n"
+
+
+def _reader_pandoc_format() -> str:
+    """Extract the pandoc `-f` value the reader uses, across every fork shape.
+
+    Forks differ: an inline "-f", "markdown..." literal, a `reader = "markdown..."`
+    variable, a `context="markdown..."` kwarg, or a PANDOC_MARKDOWN_READER constant
+    living in tools/publication_sources.py (possibly as an implicitly-joined
+    multiline string). Search both files and stitch adjacent string fragments.
+    """
+    texts = [READER.read_text(encoding="utf-8")]
+    sources = READER.parent / "publication_sources.py"
+    if sources.is_file():
+        texts.append(sources.read_text(encoding="utf-8"))
+    for src in texts:
+        for pat in (
+            r'"-f",\s*"(markdown[^"]*)"',
+            r'reader\s*=\s*"(markdown[^"]*)"',
+            r'context="(markdown[^"]*)"',
+            # constant, one or more adjacent "..." fragments (implicit join)
+            r'PANDOC_MARKDOWN_READER\s*=\s*\(?\s*((?:"[^"]*"\s*)+)\)?',
+        ):
+            m = re.search(pat, src)
+            if m:
+                raw = m.group(1)
+                # collapse implicit string concatenation into one value
+                return "".join(re.findall(r'"([^"]*)"', raw)) or raw
+    raise AssertionError("could not find the reader's pandoc -f markdown format")
+
+
+class HtmlListRenderingTests(unittest.TestCase):
+    def test_reader_declares_the_no_blank_line_list_extension(self):
+        fmt = _reader_pandoc_format()
+        self.assertIn(
+            "lists_without_preceding_blankline",
+            fmt,
+            "build_html_reader.py must enable lists_without_preceding_blankline; "
+            f"pandoc format is currently {fmt!r}. Without it, every bold-lead-in "
+            "list in the book collapses into a paragraph in the shipped HTML.",
+        )
+
+    def test_bold_lead_in_list_renders_as_ul_under_real_pandoc(self):
+        pandoc = shutil.which("pandoc")
+        if pandoc is None:
+            self.skipTest("pandoc not installed; CI installs it and runs this check")
+        fmt = _reader_pandoc_format()
+        out = subprocess.run(
+            [pandoc, "-f", fmt, "-t", "html5"],
+            input=SAMPLE,
+            capture_output=True,
+            text=True,
+            check=True,
+        ).stdout
+        self.assertIn("<ul>", out, f"list collapsed; pandoc -f {fmt} produced:\n{out}")
+        self.assertEqual(out.count("<li>"), 2, f"expected 2 <li>; got:\n{out}")
+
+
+if __name__ == "__main__":
+    unittest.main()