perf: change log_cli_level from DEBUG to WARNING

DEBUG forced all loggers to emit during tests, bypassing the cached no-op
optimization and inflating benchmark numbers. WARNING is the sensible
default; pass `-o log_cli_level=DEBUG` when debug logs are needed.

Author: fgmacedo

AGENTS.md

@@ -127,6 +127,16 @@ timeout 120 uv run pytest -n 4
 
 Testes normally run under 60s (~40s on average), so take a closer look if they take longer, it can be a regression.
 
+### Debug logging
+
+`log_cli_level` defaults to `WARNING` in `pyproject.toml`. The engine caches a no-op
+for `logger.debug` at init time — running tests with `DEBUG` would bypass this
+optimization and inflate benchmark numbers. To enable debug logs for a specific run:
+
+```bash
+uv run pytest -o log_cli_level=DEBUG tests/test_something.py
+```
+
 When analyzing warnings or extensive output, run the tests **once** saving the output to a file
 (`> /tmp/pytest-output.txt 2>&1`), then analyze the file — instead of running the suite
 repeatedly with different greps.

pyproject.toml

@@ -88,7 +88,11 @@ markers = [
 ]
 python_files = ["tests.py", "test_*.py", "*_tests.py"]
 xfail_strict = true
-log_cli_level = "DEBUG"
+# Log level WARNING by default; the engine caches a no-op for logger.debug at
+# init time, so DEBUG here would bypass that optimization and slow benchmarks.
+# To enable DEBUG logging for a specific test run:
+#   uv run pytest -o log_cli_level=DEBUG
+log_cli_level = "WARNING"
 log_cli_format = "%(relativeCreated)6.0fms %(threadName)-18s %(name)-35s %(message)s"
 log_cli_date_format = "%H:%M:%S"
 asyncio_default_fixture_loop_scope = "module"