diff --git a/tests/conftest.py b/tests/conftest.py
new file mode 100644
index 0000000..42f3c8f
--- /dev/null
+++ b/tests/conftest.py
@@ -0,0 +1,52 @@
+import sys
+import types
+import torch
+
+class _DummyProgressBar:
+    def __init__(self, total):
+        self.total = total
+        self.updated = 0
+
+    def update(self, amount):
+        self.updated += amount
+
+
+def _ensure_comfy_stubs():
+    if "comfy" in sys.modules:
+        return
+
+    comfy_module = types.ModuleType("comfy")
+
+    model_management_module = types.ModuleType("comfy.model_management")
+    model_management_module.get_torch_device = lambda: torch.device("cpu")
+    model_management_module.unet_dtype = lambda: torch.float32
+    model_management_module.load_models_gpu = lambda models: None
+
+    utils_module = types.ModuleType("comfy.utils")
+    utils_module.ProgressBar = _DummyProgressBar
+
+    samplers_module = types.ModuleType("comfy.samplers")
+
+    class _DummyKSampler:
+        def __init__(self, model, steps, device=None):
+            self.model = model
+            self.steps = steps
+            self.device = device or torch.device("cpu")
+            self.sigmas = torch.linspace(1.0, 0.0, steps)
+
+        def sample(self, *args, **kwargs):
+            return torch.zeros((1, 4, 8, 8), device=self.device)
+
+    samplers_module.KSampler = _DummyKSampler
+
+    comfy_module.model_management = model_management_module
+    comfy_module.utils = utils_module
+    comfy_module.samplers = samplers_module
+
+    sys.modules["comfy"] = comfy_module
+    sys.modules["comfy.model_management"] = model_management_module
+    sys.modules["comfy.utils"] = utils_module
+    sys.modules["comfy.samplers"] = samplers_module
+
+
+_ensure_comfy_stubs()
diff --git a/tests/test_nodes.py b/tests/test_nodes.py
new file mode 100644
index 0000000..b6fbc92
--- /dev/null
+++ b/tests/test_nodes.py
@@ -0,0 +1,102 @@
+import importlib.util
+from pathlib import Path
+
+import torch
+
+
+MODULE_PATH = Path(__file__).resolve().parents[1] / "__init__.py"
+
+
+def load_module():
+    spec = importlib.util.spec_from_file_location("flowmatching_inverter", MODULE_PATH)
+    module = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(module)
+    return module
+
+
+def test_latent_gaussian_blur_modifies_values():
+    module = load_module()
+    node = module.LatentGaussianBlur()
+    latent = {"samples": torch.randn(1, 4, 8, 8)}
+
+    (result,) = node.blur_latent(latent, sigma=1.5, blur_mode="Spatial Only")
+
+    original = latent["samples"]
+    blurred = result["samples"]
+
+    assert blurred.shape == original.shape
+    assert not torch.allclose(blurred, original)
+
+
+def test_latent_add_noise_reproducible_with_seed():
+    module = load_module()
+    node = module.LatentAddNoise()
+    latent = {"samples": torch.randn(1, 4, 8, 8)}
+
+    (first,) = node.add_noise(latent, seed=123, strength=0.8)
+    (second,) = node.add_noise(latent, seed=123, strength=0.8)
+    (third,) = node.add_noise(latent, seed=321, strength=0.8)
+
+    assert torch.allclose(first["samples"], second["samples"])
+    assert not torch.allclose(first["samples"], third["samples"])
+
+
+def test_perlin_noise_strength_affects_latent():
+    module = load_module()
+    node = module.LatentPerlinFractalNoise()
+    latent = {"samples": torch.randn(1, 4, 8, 8)}
+
+    (result,) = node.add_perlin_noise(
+        latent,
+        seed=7,
+        frequency=1.5,
+        octaves=2,
+        persistence=0.5,
+        lacunarity=2.0,
+        strength=0.5,
+        channel_mode="shared",
+    )
+
+    assert not torch.allclose(result["samples"], latent["samples"])
+
+
+def test_simplex_noise_zero_strength_is_noop():
+    module = load_module()
+    node = module.LatentSimplexNoise()
+    latent = {"samples": torch.randn(1, 4, 8, 8)}
+
+    (result,) = node.add_simplex_noise(
+        latent,
+        seed=42,
+        frequency=1.0,
+        octaves=2,
+        persistence=0.5,
+        lacunarity=2.0,
+        strength=0.0,
+        channel_mode="shared",
+        temporal_mode="locked",
+    )
+
+    assert torch.allclose(result["samples"], latent["samples"])
+
+
+def test_conditioning_nodes_modify_embeddings_and_metadata():
+    module = load_module()
+    noise_node = module.ConditioningAddNoise()
+    blur_node = module.ConditioningGaussianBlur()
+
+    embedding = torch.randn(2, 4, 6)
+    pooled = torch.randn(4)
+    conditioning = [[embedding.clone(), {"pooled_output": pooled.clone()}]]
+
+    (noised,) = noise_node.add_noise(conditioning, seed=0, strength=0.5)
+    (blurred,) = blur_node.blur(noised, sigma=1.0)
+
+    noised_embedding = noised[0][0]
+    noised_pooled = noised[0][1]["pooled_output"]
+
+    assert not torch.allclose(noised_embedding, embedding)
+    assert not torch.allclose(noised_pooled, pooled)
+
+    blurred_embedding = blurred[0][0]
+    assert blurred_embedding.shape == embedding.shape