Merge pull request #3 from amoussa1229/chore/ci-workflow-sonnet-bump

Add CI workflow and bump Sonnet to 4.6

.github/workflows/ci.yaml

@@ -0,0 +1,50 @@
+name: CI
+
+on:
+  pull_request:
+    branches: [main]
+  push:
+    branches: [main]
+
+jobs:
+  lint:
+    runs-on: ubuntu-latest
+    timeout-minutes: 5
+    steps:
+      - uses: actions/checkout@v6
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install ruff
+        run: pip install ruff
+
+      - name: Ruff check
+        run: ruff check .
+
+      - name: Ruff format check
+        run: ruff format --check .
+
+  test-collect:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    steps:
+      - uses: actions/checkout@v6
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+          cache: pip
+          cache-dependency-path: requirements.txt
+
+      - name: Install dependencies
+        run: |
+          pip install -r requirements.txt
+          pip install pytest python-dotenv
+
+      - name: Pytest collect-only
+        # Verifies all test modules + their imports resolve cleanly.
+        # Live test execution requires ANTHROPIC_API_KEY + COMPOSIO_API_KEY
+        # and runs locally before push, not in CI.
+        run: pytest --collect-only -q

.gitignore

@@ -6,3 +6,4 @@ __pycache__/
 .cache/
 .pytest_cache/
 .ruff_cache/
+.DS_Store

models.py

@@ -6,7 +6,7 @@ from langchain_google_genai import ChatGoogleGenerativeAI
 from langchain_openai import ChatOpenAI
 
 # Model IDs — single source of truth. Bump here when families ship new revs.
-CLAUDE_SONNET = "claude-sonnet-4-20250514"
+CLAUDE_SONNET = "claude-sonnet-4-6"
 CLAUDE_HAIKU = "claude-haiku-4-5-20251001"
 OPENAI_CROSS_REVIEWER = "gpt-4.1"
 GEMINI_SCANNER = "gemini-2.5-pro"

tests/test_routing_golden.py

@@ -1,8 +1,9 @@
 """Golden-set routing test for the structured router.
 
-20 labelled tasks → expected agent. Skipped if provider keys are missing
-(ANTHROPIC for the router LLM, COMPOSIO because importing the graph eagerly
-loads tools). Run with:
+20 labelled tasks → expected agent. Live tests require ANTHROPIC_API_KEY +
+COMPOSIO_API_KEY; without them the parametrized cases skip at runtime but
+the module still collects (so `pytest --collect-only` in CI works without
+secrets). Run with:
 
     pytest tests/test_routing_golden.py -v
 """
@@ -14,17 +15,12 @@ from dotenv import load_dotenv
 
 load_dotenv(os.path.join(os.path.dirname(__file__), "..", ".env"))
 
-if not os.getenv("ANTHROPIC_API_KEY"):
-    pytest.skip(
-        "ANTHROPIC_API_KEY not set; skipping live router tests.",
-        allow_module_level=True,
-    )
-if not os.getenv("COMPOSIO_API_KEY"):
-    pytest.skip(
-        "COMPOSIO_API_KEY not set; skipping live router tests.", allow_module_level=True
-    )
-
-from graph import router_node  # noqa: E402
+_REQUIRED_KEYS = ("ANTHROPIC_API_KEY", "COMPOSIO_API_KEY")
+_missing = [k for k in _REQUIRED_KEYS if not os.getenv(k)]
+requires_keys = pytest.mark.skipif(
+    bool(_missing),
+    reason=f"Requires {', '.join(_REQUIRED_KEYS)}; missing: {', '.join(_missing)}",
+)
 
 
 GOLDEN_SET: list[tuple[str, str]] = [
@@ -107,8 +103,13 @@ def test_golden_set_size():
     assert len(GOLDEN_SET) == 20
 
 
+@requires_keys
 @pytest.mark.parametrize("task,expected", GOLDEN_SET)
 def test_router_picks_expected_agent(task: str, expected: str):
+    # Lazy import — graph.py loads composio_tools at module import and needs
+    # COMPOSIO_API_KEY. Keep the import inside the skipif-guarded test body.
+    from graph import router_node
+
     out = router_node({"task": task, "messages": []})
     assert out["route"] == expected, (
         f"task={task!r} got={out['route']} expected={expected}"