diff --git a/.github/workflows/tests.yml b/.github/workflows/tests.yml
new file mode 100644
index 0000000..125217d
--- /dev/null
+++ b/.github/workflows/tests.yml
@@ -0,0 +1,22 @@
+name: CI
+
+on:
+  pull_request:
+    branches: [main]
+  push:
+    branches: [main]
+
+jobs:
+  run-tests:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v5
+
+      - name: Set up Python
+        run: uv python install 3.12
+
+      - name: Run tests
+        run: uv run --extra dev pytest tests/ -v
diff --git a/tests/test_channel_theme.py b/tests/test_channel_theme.py
index 3d22a39..7ccfd7b 100644
--- a/tests/test_channel_theme.py
+++ b/tests/test_channel_theme.py
@@ -41,22 +41,19 @@ def test_colorfgbg_light(self):
 
     def test_default_is_dark(self):
         with patch.dict("os.environ", {"ANTON_THEME": "", "COLORFGBG": ""}, clear=False):
-            with patch("anton.channel.theme.os.uname") as mock_uname:
-                mock_uname.return_value.sysname = "Linux"
+            with patch("anton.channel.theme.sys.platform", "linux"):
                 assert detect_color_mode() == "dark"
 
     def test_macos_dark_mode(self):
         with patch.dict("os.environ", {"ANTON_THEME": "", "COLORFGBG": ""}, clear=False):
-            with patch("anton.channel.theme.os.uname") as mock_uname:
-                mock_uname.return_value.sysname = "Darwin"
+            with patch("anton.channel.theme.sys.platform", "darwin"):
                 with patch("anton.channel.theme.subprocess.run") as mock_run:
                     mock_run.return_value.returncode = 0
                     assert detect_color_mode() == "dark"
 
     def test_macos_light_mode(self):
         with patch.dict("os.environ", {"ANTON_THEME": "", "COLORFGBG": ""}, clear=False):
-            with patch("anton.channel.theme.os.uname") as mock_uname:
-                mock_uname.return_value.sysname = "Darwin"
+            with patch("anton.channel.theme.sys.platform", "darwin"):
                 with patch("anton.channel.theme.subprocess.run") as mock_run:
                     mock_run.return_value.returncode = 1
                     assert detect_color_mode() == "light"
diff --git a/tests/test_chat_scratchpad.py b/tests/test_chat_scratchpad.py
index 94d0d0a..645addf 100644
--- a/tests/test_chat_scratchpad.py
+++ b/tests/test_chat_scratchpad.py
@@ -205,6 +205,7 @@ async def test_scratchpad_dump_streams_tool_result(self):
         """dump action yields a StreamToolResult for display, but sends a short
         summary back to the LLM to avoid it parroting the full notebook."""
         mock_llm = AsyncMock()
+        mock_llm.plan = AsyncMock(return_value=_text_response("STATUS: COMPLETE — task done"))
 
         call_count = 0
 
@@ -259,6 +260,7 @@ async def test_scratchpad_in_streaming_path(self):
         final_response = _text_response("Got 99.")
 
         mock_llm = AsyncMock()
+        mock_llm.plan = AsyncMock(return_value=_text_response("STATUS: COMPLETE — task done"))
 
         call_count = 0
 
diff --git a/tests/test_chat_ui.py b/tests/test_chat_ui.py
index 471f9bf..68045bd 100644
--- a/tests/test_chat_ui.py
+++ b/tests/test_chat_ui.py
@@ -226,4 +226,3 @@ def test_text_routes_to_buffer_after_tools(self, MockLive):
         assert display._initial_text == "Initial text"
         assert display._buffer == "Answer text"
         assert display._in_tool_phase
-        assert display._answer_started
diff --git a/tests/test_datasource.py b/tests/test_datasource.py
index 3989cdf..dc7a13a 100644
--- a/tests/test_datasource.py
+++ b/tests/test_datasource.py
@@ -583,7 +583,7 @@ async def test_partial_save_on_n_answer(self, registry, vault_dir, make_session)
         conns = vault.list_connections()
         assert len(conns) == 1
         assert conns[0]["engine"] == "postgresql"
-        assert conns[0]["name"].isdigit()
+        assert conns[0]["name"].isalnum()
         session._scratchpads.get_or_create.assert_not_called()
 
     @pytest.mark.asyncio
diff --git a/tests/test_settings.py b/tests/test_settings.py
index 5833fe8..ebc318a 100644
--- a/tests/test_settings.py
+++ b/tests/test_settings.py
@@ -8,21 +8,37 @@
 from anton.config.settings import AntonSettings
 
 
+_ANTON_MODEL_KEYS = [
+    "ANTON_PLANNING_PROVIDER",
+    "ANTON_PLANNING_MODEL",
+    "ANTON_CODING_PROVIDER",
+    "ANTON_CODING_MODEL",
+]
+
+
 class TestAntonSettingsDefaults:
-    def test_default_planning_provider(self):
-        s = AntonSettings(anthropic_api_key="test")
+    def test_default_planning_provider(self, monkeypatch):
+        for k in _ANTON_MODEL_KEYS:
+            monkeypatch.delenv(k, raising=False)
+        s = AntonSettings(anthropic_api_key="test", _env_file=None)
         assert s.planning_provider == "anthropic"
 
-    def test_default_planning_model(self):
-        s = AntonSettings(anthropic_api_key="test")
+    def test_default_planning_model(self, monkeypatch):
+        for k in _ANTON_MODEL_KEYS:
+            monkeypatch.delenv(k, raising=False)
+        s = AntonSettings(anthropic_api_key="test", _env_file=None)
         assert s.planning_model == "claude-sonnet-4-6"
 
-    def test_default_coding_provider(self):
-        s = AntonSettings(anthropic_api_key="test")
+    def test_default_coding_provider(self, monkeypatch):
+        for k in _ANTON_MODEL_KEYS:
+            monkeypatch.delenv(k, raising=False)
+        s = AntonSettings(anthropic_api_key="test", _env_file=None)
         assert s.coding_provider == "anthropic"
 
-    def test_default_coding_model(self):
-        s = AntonSettings(anthropic_api_key="test")
+    def test_default_coding_model(self, monkeypatch):
+        for k in _ANTON_MODEL_KEYS:
+            monkeypatch.delenv(k, raising=False)
+        s = AntonSettings(anthropic_api_key="test", _env_file=None)
         assert s.coding_model == "claude-haiku-4-5-20251001"
 
     def test_default_memory_dir(self):