diff --git a/README.md b/README.md
index 68d076e..a5b3a79 100644
--- a/README.md
+++ b/README.md
@@ -19,6 +19,7 @@ PromptLens runs golden test sets against multiple models, scores outputs using L
 - **Multiple Export Formats** - HTML, JSON, CSV, and Markdown outputs
 - **Parallel Execution** - Async execution with configurable concurrency and retry logic
 - **Portable & Local** - No cloud backend, all data stays on your machine
+- **Safer HTTP Integrations** - Unparseable HTTP provider responses fail fast with clear error context
 - **Easy to Extend** - Plugin architecture for custom providers, judges, and exporters
 
 ---
diff --git a/promptlens/providers/http.py b/promptlens/providers/http.py
index f1828e2..1bbaab9 100644
--- a/promptlens/providers/http.py
+++ b/promptlens/providers/http.py
@@ -123,6 +123,11 @@ async def _make_request() -> ModelResponse:
 
                 # Extract content (try common response formats)
                 content = self._extract_content(data)
+                if not content.strip():
+                    raise ValueError(
+                        "HTTP provider response did not contain parseable text content. "
+                        f"Top-level keys: {sorted(data.keys()) if isinstance(data, dict) else type(data).__name__}"
+                    )
 
                 # Local models typically don't provide token counts or cost
                 return ModelResponse(
diff --git a/tests/test_http_provider_empty_content_guard.py b/tests/test_http_provider_empty_content_guard.py
new file mode 100644
index 0000000..19199e8
--- /dev/null
+++ b/tests/test_http_provider_empty_content_guard.py
@@ -0,0 +1,51 @@
+from unittest.mock import patch
+
+import pytest
+
+from promptlens.models.config import ProviderConfig
+from promptlens.providers.http import HTTPProvider
+
+
+def _provider() -> HTTPProvider:
+    return HTTPProvider(
+        ProviderConfig(
+            name="http",
+            model="test-model",
+            endpoint="http://localhost:11434/api/generate",
+        )
+    )
+
+
+@pytest.mark.asyncio
+async def test_generate_returns_error_when_content_unparseable() -> None:
+    provider = _provider()
+
+    class _FakeResponse:
+        def raise_for_status(self) -> None:
+            return None
+
+        async def json(self):
+            return {"foo": "bar"}
+
+        async def __aenter__(self):
+            return self
+
+        async def __aexit__(self, exc_type, exc, tb):
+            return None
+
+    class _FakeSession:
+        def post(self, *args, **kwargs):
+            return _FakeResponse()
+
+        async def __aenter__(self):
+            return self
+
+        async def __aexit__(self, exc_type, exc, tb):
+            return None
+
+    with patch("promptlens.providers.http.aiohttp.ClientSession", return_value=_FakeSession()):
+        result = await provider.generate("hello")
+
+    assert result.error is not None
+    assert "did not contain parseable text content" in result.error
+    assert result.content == ""