diff --git a/src/skillspector/input_handler.py b/src/skillspector/input_handler.py
index 21c83b71e..767315f66 100644
--- a/src/skillspector/input_handler.py
+++ b/src/skillspector/input_handler.py
@@ -813,7 +813,7 @@ def resolve(self, input_path: str) -> tuple[Path, str]:
         if self._is_file_url(input_path):
             return self._download_file(input_path), "url"
         normalized_local_path = validate_local_input_path(Path(input_path))
-        if input_path.endswith(".zip"):
+        if input_path.lower().endswith(".zip"):
             return self._extract_zip(normalized_local_path), "zip"
         if input_path.endswith(".md"):
             return self._wrap_single_file(normalized_local_path), "file"
@@ -1366,7 +1366,7 @@ def _download_file(self, url: str) -> Path:
             download_path.unlink(missing_ok=True)
             raise
 
-        is_zip = filename.endswith(".zip") or content_type.startswith("application/zip")
+        is_zip = filename.lower().endswith(".zip") or content_type.startswith("application/zip")
         if is_zip:
             zip_path = temp_dir / "download.zip"
             download_path.replace(zip_path)
@@ -1389,7 +1389,7 @@ def _download_transitive_file(self, url: str) -> Path:
             raise ValueError(f"Failed to download file: {exc}") from exc
         temp_dir = self._get_temp_dir()
         self._record_artifacts(1)
-        if filename.endswith(".zip") or headers.get("content-type", "").startswith(
+        if filename.lower().endswith(".zip") or headers.get("content-type", "").startswith(
             "application/zip"
         ):
             zip_path = temp_dir / "download.zip"
diff --git a/tests/unit/test_input_handler.py b/tests/unit/test_input_handler.py
index 663274b12..dbd3abe9b 100644
--- a/tests/unit/test_input_handler.py
+++ b/tests/unit/test_input_handler.py
@@ -374,12 +374,13 @@ def test_windows_no_follow_open_rejects_canonical_path_mismatch(
         _open_regular_file_from_windows_handle(source)
 
 
-def test_resolve_zip_file(tmp_path: Path) -> None:
-    """Resolving a .zip file extracts and returns the extract dir."""
+@pytest.mark.parametrize("extension", [".zip", ".ZIP", ".ZiP"])
+def test_resolve_zip_file(tmp_path: Path, extension: str) -> None:
+    """ZIP extensions are recognized regardless of case for local inputs."""
     import zipfile
 
     (tmp_path / "SKILL.md").write_text("# Skill", encoding="utf-8")
-    zip_path = tmp_path / "skill.zip"
+    zip_path = tmp_path / f"skill{extension}"
     with zipfile.ZipFile(zip_path, "w") as zf:
         zf.write(tmp_path / "SKILL.md", "SKILL.md")
     handler = InputHandler()
@@ -387,6 +388,7 @@ def test_resolve_zip_file(tmp_path: Path) -> None:
         resolved, source_type = handler.resolve(str(zip_path))
         assert resolved.is_dir()
         assert source_type == "zip"
+        assert (resolved / "SKILL.md").read_text(encoding="utf-8") == "# Skill"
     finally:
         handler.cleanup()
 
@@ -601,6 +603,45 @@ def serve(request: httpx.Request) -> httpx.Response:
         handler.cleanup()
 
 
+@pytest.mark.parametrize("budgeted", [False, True], ids=["direct", "workflow-budget"])
+@pytest.mark.parametrize("extension", [".zip", ".ZIP", ".ZiP"])
+def test_file_url_download_extracts_zip_regardless_of_extension_case(
+    monkeypatch: pytest.MonkeyPatch, extension: str, budgeted: bool
+) -> None:
+    """ZIP filenames trigger extraction even without a ZIP content type."""
+    import io
+    import zipfile
+
+    skill = b"---\nname: demo\ndescription: demo\n---\n# Demo\n"
+    archive = io.BytesIO()
+    with zipfile.ZipFile(archive, "w") as zf:
+        zf.writestr("SKILL.md", skill)
+    url = f"https://raw.githubusercontent.com/org/repo/main/skill{extension}"
+    requested: list[str] = []
+
+    def serve(request: httpx.Request) -> httpx.Response:
+        requested.append(str(request.url))
+        return httpx.Response(
+            200, content=archive.getvalue(), headers={"content-type": "application/octet-stream"}
+        )
+
+    real_client = httpx.Client
+    monkeypatch.setattr(
+        "skillspector.input_handler.httpx.Client",
+        lambda *args, **kwargs: real_client(*args, transport=httpx.MockTransport(serve), **kwargs),
+    )
+    monkeypatch.setattr("skillspector.input_handler._is_private_ip", lambda _host: False)
+    handler = InputHandler(transitive_budget=WorkflowResourceBudget() if budgeted else None)
+    try:
+        resolved, source_type = handler.resolve(url)
+
+        assert source_type == "url"
+        assert requested == [url]
+        assert (resolved / "SKILL.md").read_bytes() == skill
+    finally:
+        handler.cleanup()
+
+
 def test_validate_url_host_scp_extracts_github() -> None:
     """_validate_url_host extracts 'github.com' from an scp-style URL."""
     with patch("skillspector.input_handler._is_private_ip", return_value=False):