Skip to content

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

fix: integrate B6+B7 — fix cross-test contamination + injector mock parity

chopratejas2fb905fdb016
diff --git a/tests/test_memory_tool_mode.py b/tests/test_memory_tool_mode.py
index 6f5e86237..b87c2217f 100644
--- a/tests/test_memory_tool_mode.py
+++ b/tests/test_memory_tool_mode.py
@@ -108,13 +108,27 @@ def test_tool_mode_skip_emits_structured_log(caplog: Any) -> None:
Realignment build constraint: every cache-affecting decision is logged
in the ``event=foo key=val`` style so operators can audit routing.
+
+ NOTE: caplog captures at the root logger via propagation. When other
+ tests in the suite trigger proxy startup, ``_setup_file_logging`` sets
+ ``headroom.propagate=False`` and attaches a file handler. The conftest
+ autouse reset is fragile against fixture ordering, so we attach
+ ``caplog.handler`` directly to the target logger here. That way the
+ capture works regardless of propagation state.
"""
handler, _backend = _build_tool_mode_handler()
- with caplog.at_level(logging.INFO, logger="headroom.proxy.memory_handler"):
+ target_logger = logging.getLogger("headroom.proxy.memory_handler")
+ previous_level = target_logger.level
+ target_logger.setLevel(logging.INFO)
+ target_logger.addHandler(caplog.handler)
+ try:
result = asyncio.run(
handler.search_and_format_context("alpha", [{"role": "user", "content": "hi"}])
)
+ finally:
+ target_logger.removeHandler(caplog.handler)
+ target_logger.setLevel(previous_level)
assert result is None
skip_records = [r for r in caplog.records if "event=memory_mode_skip" in r.getMessage()]
diff --git a/tests/test_proxy_anthropic_cache_stability.py b/tests/test_proxy_anthropic_cache_stability.py
index 6daa01c1b..a085c326b 100644
--- a/tests/test_proxy_anthropic_cache_stability.py
+++ b/tests/test_proxy_anthropic_cache_stability.py
@@ -459,6 +459,9 @@ def test_ccr_system_instruction_injection_disabled_when_prefix_frozen(monkeypatc
def process_request(self, messages, tools): # noqa: ANN001
return messages, tools, False
+ def scan_for_markers(self, messages): # noqa: ANN001
+ return []
+
monkeypatch.setattr("headroom.ccr.CCRToolInjector", _FakeInjector)
async def _fake_retry(method, url, headers, body, stream=False, **kwargs): # noqa: ANN001
@@ -523,6 +526,9 @@ def test_ccr_tool_injection_disabled_when_prefix_frozen(monkeypatch) -> None:
def process_request(self, messages, tools): # noqa: ANN001
return messages, tools, False
+ def scan_for_markers(self, messages): # noqa: ANN001
+ return []
+
monkeypatch.setattr("headroom.ccr.CCRToolInjector", _FakeInjector)
async def _fake_retry(method, url, headers, body, stream=False, **kwargs): # noqa: ANN001