fix: stop eval modules leaking .env into the test process (#1482)

evals/reviewer/{target,run_eval,build_dataset} called load_dotenv() at
import time, so importing them in tests injected the real .env (live
LANGGRAPH_URL, tokens) into the whole pytest process. The slack-context
default-repo tests then reached the real LangGraph store via
get_team_default_repo() and picked up the developer's actual team
default repo, failing in full-suite runs while passing in isolation.

Move load_dotenv() into the CLI entrypoints (all env reads were already
lazy), and patch get_team_default_repo in the two affected tests so they
stay hermetic regardless of environment.

Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>

evals/reviewer/build_dataset.py

@@ -21,8 +21,6 @@ from pathlib import Path
 from dotenv import load_dotenv
 from langsmith import Client
 
-load_dotenv()
-
 GOLDENS_DIR = Path(__file__).parent / "golden_comments"
 PR_URL_RE = re.compile(r"github\.com/([^/]+)/([^/]+)/pull/(\d+)")
 
@@ -115,6 +113,7 @@ def upload(dataset_name: str, examples: list[dict]) -> None:
 
 
 def main() -> None:
+    load_dotenv()
     ap = argparse.ArgumentParser()
     ap.add_argument("--dataset-name", default="openswe-reviewer-v1")
     ap.add_argument("--dry-run", action="store_true", help="Build examples but don't upload.")

evals/reviewer/run_eval.py

@@ -22,8 +22,6 @@ from langsmith.schemas import Example
 from evals.reviewer.judge import aggregate_pr, judge_match
 from evals.reviewer.target import drain_thread_ids, get_langgraph_url, review_pr
 
-load_dotenv()
-
 logger = logging.getLogger(__name__)
 
 CONFIG_PATH = Path(__file__).with_name("config.toml")
@@ -127,6 +125,7 @@ async def _cleanup_threads(thread_ids: Iterable[str]) -> None:
 
 
 async def main() -> None:
+    load_dotenv()
     config = _load_config()
     _apply_config_to_env(config)
 

evals/reviewer/target.py

@@ -13,13 +13,10 @@ import os
 import threading
 from typing import Any, Literal, cast
 
-from dotenv import load_dotenv
 from langgraph_sdk import get_client
 
 from agent.reviewer_findings import Finding, Severity, filter_findings_for_publish
 
-load_dotenv()
-
 DEFAULT_REVIEWER_ASSISTANT_ID = "reviewer"
 DEFAULT_LANGGRAPH_URL = "http://localhost:2024"
 ScoreMode = Literal["all_findings", "surfaced_findings"]

tests/test_slack_context.py

@@ -323,12 +323,17 @@ def test_get_slack_repo_config_uses_existing_thread_repo(
     assert not posted
 
 
+async def _no_team_default_repo() -> dict[str, str] | None:
+    return None
+
+
 def test_get_slack_repo_config_new_thread_uses_default(
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
     threads_client = _FakeThreadsClient(raise_not_found=True)
     monkeypatch.setattr(webapp, "SLACK_REPO_OWNER", "default-owner")
     monkeypatch.setattr(webapp, "SLACK_REPO_NAME", "default-repo")
+    monkeypatch.setattr(webapp, "get_team_default_repo", _no_team_default_repo)
 
     monkeypatch.setattr(webapp, "get_client", lambda url: _FakeClient(threads_client))
 
@@ -343,6 +348,7 @@ def test_get_slack_repo_config_existing_thread_without_repo_uses_default(
     threads_client = _FakeThreadsClient(thread={"metadata": {}})
     monkeypatch.setattr(webapp, "SLACK_REPO_OWNER", "default-owner")
     monkeypatch.setattr(webapp, "SLACK_REPO_NAME", "default-repo")
+    monkeypatch.setattr(webapp, "get_team_default_repo", _no_team_default_repo)
 
     monkeypatch.setattr(webapp, "get_client", lambda url: _FakeClient(threads_client))