diff --git a/.github/workflows/checks.yaml b/.github/workflows/checks.yaml
index 4d4b1812..e3f92a6a 100644
--- a/.github/workflows/checks.yaml
+++ b/.github/workflows/checks.yaml
@@ -80,6 +80,21 @@ jobs:
         run: |
           mise exec python@${{ matrix.python-version }} -- python ./py/scripts/nox-matrix.py ${{ matrix.shard }} 4
 
+  vulture:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    steps:
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4.3.1
+      - name: Set up mise
+        uses: jdx/mise-action@1648a7812b9aeae629881980618f079932869151 # v4.0.1
+        with:
+          cache: true
+          experimental: true
+      - name: Install dev dependencies
+        run: mise exec -- make -C py install-dev
+      - name: Run vulture dead-code check
+        run: mise exec -- make -C py vulture
+
   adk-py:
     uses: ./.github/workflows/adk-py-test.yaml
 
@@ -116,6 +131,7 @@ jobs:
       - ensure-pinned-actions
       - smoke
       - nox
+      - vulture
       - adk-py
       - langchain-py
       - upload-wheel
@@ -141,6 +157,7 @@ jobs:
           check_result "ensure-pinned-actions" "${{ needs.ensure-pinned-actions.result }}"
           check_result "smoke" "${{ needs.smoke.result }}"
           check_result "nox" "${{ needs.nox.result }}"
+          check_result "vulture" "${{ needs.vulture.result }}"
           check_result "adk-py" "${{ needs.adk-py.result }}"
           check_result "langchain-py" "${{ needs.langchain-py.result }}"
           check_result "upload-wheel" "${{ needs.upload-wheel.result }}"
diff --git a/py/Makefile b/py/Makefile
index 56becdfe..5677f3a0 100644
--- a/py/Makefile
+++ b/py/Makefile
@@ -2,7 +2,7 @@ PYTHON ?= python
 UV := $(PYTHON) -m uv
 UV_VERSION := $(shell awk '$$1=="uv" { print $$2 }' ../.tool-versions)
 
-.PHONY: lint pylint test test-wheel _template-version clean fixup build verify-build verify help install-build-deps install-dev install-optional test-core _check-git-clean bench bench-compare
+.PHONY: lint pylint vulture test test-wheel _template-version clean fixup build verify-build verify help install-build-deps install-dev install-optional test-core _check-git-clean bench bench-compare
 
 clean:
 	rm -rf build dist
@@ -17,6 +17,9 @@ lint: fixup
 pylint:
 	nox -s pylint
 
+vulture:
+	$(PYTHON) -m vulture src/braintrust --min-confidence 100 --exclude "*/test_*.py,*/_test_*.py"
+
 test:
 	nox -x
 
@@ -58,7 +61,7 @@ _check-git-clean:
 
 verify-build: _check-git-clean build test-wheel
 
-verify: lint test
+verify: lint vulture test
 
 install-build-deps:
 	$(if $(UV_VERSION),,$(error Failed to read uv version from ../.tool-versions))
@@ -85,6 +88,7 @@ help:
 	@echo "  install-dev         - Install package in development mode with all dependencies"
 	@echo "  lint                - Run pylint checks"
 	@echo "  pylint              - Run pylint without pre-commit hooks"
+	@echo "  vulture             - Detect unused/dead code"
 	@echo "  test                - Run all tests"
 	@echo "  test-core           - Run core tests only"
 	@echo "  test-wheel          - Run tests against built wheel"
diff --git a/py/requirements-dev.txt b/py/requirements-dev.txt
index 14fb3c47..04799aaf 100644
--- a/py/requirements-dev.txt
+++ b/py/requirements-dev.txt
@@ -5,6 +5,7 @@ pre-commit==4.5.1
 pydoc-markdown==4.8.2
 pylint==4.0.5
 pyperf==2.10.0
+vulture==2.16
 pytest==9.0.2
 pytest-asyncio==1.3.0
 pytest-forked==1.6.0
diff --git a/py/src/braintrust/framework.py b/py/src/braintrust/framework.py
index 378aa852..5aa9172a 100644
--- a/py/src/braintrust/framework.py
+++ b/py/src/braintrust/framework.py
@@ -203,7 +203,7 @@ def tags(self) -> Sequence[str]:
         """
 
     @abc.abstractmethod
-    def report_progress(self, progress: TaskProgressEvent) -> None:
+    def report_progress(self, progress: TaskProgressEvent) -> None:  # noqa: F841
         """
         Report progress that will show up in the playground.
         """
@@ -459,7 +459,7 @@ class EvalResultWithSummary(SerializableDataClass, Generic[Input, Output]):
     summary: ExperimentSummary
     results: list[EvalResult[Input, Output]]
 
-    def _repr_pretty_(self, p, cycle):
+    def _repr_pretty_(self, p, _cycle):
         p.text(f'EvalResultWithSummary(summary="...", results=[...])')
 
 
diff --git a/py/src/braintrust/integrations/base.py b/py/src/braintrust/integrations/base.py
index 68d2936a..21094fa7 100644
--- a/py/src/braintrust/integrations/base.py
+++ b/py/src/braintrust/integrations/base.py
@@ -194,7 +194,7 @@ def applies(cls, module: Any | None, version: str | None, *, target: Any | None
 
     @classmethod
     @abstractmethod
-    def patch_class(cls, target_class: type[Any]) -> bool | None:
+    def patch_class(cls, target_class: type[Any]) -> bool | None:  # noqa: F841
         """Patch one discovered class.
 
         Return ``False`` to skip marking the class as patched. Any other return
diff --git a/py/src/braintrust/integrations/langchain/callbacks.py b/py/src/braintrust/integrations/langchain/callbacks.py
index 50650da0..1c553742 100644
--- a/py/src/braintrust/integrations/langchain/callbacks.py
+++ b/py/src/braintrust/integrations/langchain/callbacks.py
@@ -552,7 +552,7 @@ def on_text(
 
     def on_retry(
         self,
-        retry_state: RetryCallState,
+        _retry_state: RetryCallState,
         *,
         run_id: UUID,
         parent_run_id: UUID | None = None,
diff --git a/py/src/braintrust/logger.py b/py/src/braintrust/logger.py
index f7c56d2b..ef7095ac 100644
--- a/py/src/braintrust/logger.py
+++ b/py/src/braintrust/logger.py
@@ -1445,7 +1445,7 @@ def _register_dropped_item_count(self, num_items):
                 self._queue_drop_logging_state["last_logged_timestamp"] = time_now
 
     @staticmethod
-    def _write_payload_to_dir(payload_dir, payload, debug_logging_adjective=None):
+    def _write_payload_to_dir(payload_dir, payload):
         payload_file = os.path.join(payload_dir, f"payload_{time.time()}_{str(uuid.uuid4())[:8]}.json")
         try:
             os.makedirs(payload_dir, exist_ok=True)
@@ -2839,7 +2839,7 @@ def _validate_and_sanitize_experiment_log_partial_args(event: Mapping[str, Any])
 # Note that this only checks properties that are expected of a complete event.
 # _validate_and_sanitize_experiment_log_partial_args should still be invoked
 # (after handling special fields like 'id').
-def _validate_and_sanitize_experiment_log_full_args(event: Mapping[str, Any], has_dataset: bool) -> Mapping[str, Any]:
+def _validate_and_sanitize_experiment_log_full_args(event: Mapping[str, Any]) -> Mapping[str, Any]:
     input = event.get("input")
     inputs = event.get("inputs")
     if (input is not None and inputs is not None) or (input is None and inputs is None):
@@ -3868,8 +3868,7 @@ def log(
                 metadata=metadata,
                 metrics=metrics,
                 id=id,
-            ),
-            self.dataset is not None,
+            )
         )
         span = self._start_span_impl(start_time=self.last_start_time, lookup_span_parent=False, **event)
         self.last_start_time = span.end()