diff --git a/tests/test_graceful_shutdown.py b/tests/test_graceful_shutdown.py
index 7bff1e5a1..24358295d 100644
--- a/tests/test_graceful_shutdown.py
+++ b/tests/test_graceful_shutdown.py
@@ -109,9 +109,40 @@ def test_run_server_installs_cancelled_error_filter(monkeypatch: pytest.MonkeyPa
 
     from headroom.proxy.server import run_server
 
-    run_server(ProxyConfig(), print_banner=False)
+    # `uvicorn.error` is a process-global logger and run_server only installs
+    # the filter when one is not already attached. Any earlier test in the
+    # session that reached run_server therefore leaves it installed, and this
+    # test would observe zero installs. Isolate the global state rather than
+    # depend on test ordering.
+    uvicorn_error_logger = logging.getLogger("uvicorn.error")
+    preexisting = [
+        item
+        for item in uvicorn_error_logger.filters
+        if isinstance(item, _SuppressCancelledErrorFilter)
+    ]
+    for item in preexisting:
+        uvicorn_error_logger.removeFilter(item)
 
-    assert len(installed_filters) == 1, "Expected exactly one _SuppressCancelledErrorFilter"
+    try:
+        run_server(ProxyConfig(), print_banner=False)
+
+        assert len(installed_filters) == 1, "Expected exactly one _SuppressCancelledErrorFilter"
+
+        # The idempotence guard is the real contract: a second call must not
+        # stack a duplicate filter on the shared logger.
+        run_server(ProxyConfig(), print_banner=False)
+        attached = [
+            item
+            for item in uvicorn_error_logger.filters
+            if isinstance(item, _SuppressCancelledErrorFilter)
+        ]
+        assert len(attached) == 1, f"filter stacked on repeat calls: {len(attached)}"
+    finally:
+        for item in list(uvicorn_error_logger.filters):
+            if isinstance(item, _SuppressCancelledErrorFilter):
+                uvicorn_error_logger.removeFilter(item)
+        for item in preexisting:
+            original_add_filter(uvicorn_error_logger, item)
 
 
 # ---------------------------------------------------------------------------
diff --git a/tests/test_learn/test_integration.py b/tests/test_learn/test_integration.py
index 3ed01d189..be465a612 100644
--- a/tests/test_learn/test_integration.py
+++ b/tests/test_learn/test_integration.py
@@ -294,9 +294,20 @@ class TestCodexIntegration:
 
         assert len(sessions) > 0
 
-        # Codex has only Bash tool (shell)
+        # This runs against whatever Codex sessions the machine actually has,
+        # so it must not pin one release's tool vocabulary. Codex has renamed
+        # its shell tool across versions (`Bash` -> `shell` -> `exec`) and
+        # 0.149.0 added agent tools alongside it, which is what the pipeline
+        # has to keep parsing.
         all_tools = {tc.name for s in sessions for tc in s.tool_calls}
-        assert "Bash" in all_tools
+        assert all_tools, "pipeline extracted no tool calls from real Codex sessions"
+
+        shell_tool_aliases = {"Bash", "shell", "exec", "local_shell", "container.exec"}
+        assert all_tools & shell_tool_aliases, (
+            "no shell-execution tool recognised in real Codex sessions; "
+            f"saw {sorted(all_tools)} -- if Codex renamed it again, add the "
+            "new name to shell_tool_aliases"
+        )
 
     def test_codex_writer_targets_agents_md(self, tmp_path):
         """Codex writer should target AGENTS.md, not CLAUDE.md."""
diff --git a/tests/test_release_workflows.py b/tests/test_release_workflows.py
index b3b3a9c07..732edda23 100644
--- a/tests/test_release_workflows.py
+++ b/tests/test_release_workflows.py
@@ -339,23 +339,35 @@ def test_no_native_tls_in_wheel_build_tree() -> None:
     import subprocess
 
     for crate in ("headroom-py", "headroom-proxy", "headroom-core"):
-        result = subprocess.run(
-            [
-                "cargo",
-                "tree",
-                "--target",
-                "x86_64-unknown-linux-gnu",
-                "-p",
-                crate,
-                "-i",
-                "native-tls",
-            ],
-            cwd=str(ROOT),
-            capture_output=True,
-            text=True,
-            check=False,
-        )
+        try:
+            result = subprocess.run(
+                [
+                    "cargo",
+                    "tree",
+                    "--target",
+                    "x86_64-unknown-linux-gnu",
+                    "-p",
+                    crate,
+                    "-i",
+                    "native-tls",
+                ],
+                cwd=str(ROOT),
+                capture_output=True,
+                text=True,
+                check=False,
+            )
+        except FileNotFoundError:
+            pytest.skip("cargo is unavailable in this environment")
+        # Same environment gates as the openssl-sys dual above: a cargo
+        # failure that is not "package did not match" means the Linux wheel
+        # target is unavailable here, not that native-tls came back.
         not_in_tree = result.returncode != 0 and "did not match any packages" in result.stderr
+        if result.returncode != 0 and "package ID specification `native-tls` did not match" not in (
+            result.stderr + result.stdout
+        ):
+            pytest.skip(
+                "cargo dependency tree for the Linux wheel target is unavailable in this environment"
+            )
         assert not_in_tree, (
             f"native-tls is back in {crate}'s build tree — likely some "
             f"crate's `default-features = true` re-enabled native-tls "