tests/unit/models/config/test_quota_scheduler_config.py (1)

10-17: Test correctly validates default values for new fields.

Good coverage for the new database_reconnection_count and database_reconnection_delay fields with their expected defaults.

However, the pyright: ignore[reportCallIssue] on line 12 is inconsistent with test_quota_scheduler_custom_configuration (lines 28-32) which doesn't have the ignore comment despite both being valid constructor calls. If the ignore is truly needed due to ConfigurationBase or Pydantic typing quirks, consider adding it consistently; otherwise, remove it from here.

tests/unit/models/config/test_quota_handlers_config.py (1)

12-24: Consider adding assertions for scheduler field values.

The test passes explicit values to QuotaSchedulerConfiguration but only asserts cfg.scheduler is not None. For completeness, consider verifying the scheduler's field values:

🔧 Suggested enhancement

     assert cfg.scheduler is not None
+    assert cfg.scheduler.period == 10
+    assert cfg.scheduler.database_reconnection_count == 10
+    assert cfg.scheduler.database_reconnection_delay == 60
     assert not cfg.enable_token_history

tests/unit/models/config/test_user_data_collection.py (3)

9-17: Test name does not reflect the scenario being tested.

The function name test_user_data_collection_feedback_enabled suggests it tests behavior when feedback is enabled, but it actually tests feedback_enabled=False. Similarly, test_user_data_collection_feedback_disabled tests the error case when feedback is enabled. Consider renaming for clarity:

• test_user_data_collection_feedback_enabled → test_user_data_collection_feedback_disabled_valid
• test_user_data_collection_feedback_disabled → test_user_data_collection_feedback_enabled_missing_storage

Same pattern applies to the transcripts tests.

---

75-89: Test assumes /root is non-writable.

This test relies on /root being non-writable, which holds in typical Unix environments but would fail if tests run as root (e.g., some CI containers). Consider using pytest's tmp_path fixture with explicit permission manipulation for a more robust test.

---

12-14: Consider addressing pyright/Pydantic compatibility at project level.

Multiple # pyright: ignore[reportCallIssue] comments suppress type errors caused by pyright not fully understanding Pydantic's Field() defaults. While this workaround is acceptable, consider:

1. Using pylance or pyright Pydantic plugin for better type inference
2. Documenting this pattern in the project's contributing guidelines if it's expected across test files

tests/unit/models/config/test_conversation_history.py (1)

25-25: Consider using a centralized pyright configuration for Pydantic models.

The widespread # pyright: ignore[reportCallIssue] comments throughout the test files indicate that pyright doesn't fully understand Pydantic's optional field semantics. While the inline suppressions work, you might consider:

1. Adding pyrightconfig.json with Pydantic support configuration
2. Using pydantic.v1 or ensuring the pyright Pydantic plugin is enabled

This would reduce the need for repetitive inline comments across test files. However, the current approach is functional and acceptable for test code.

tests/unit/models/config/test_postgresql_database_configuration.py (1)

18-23: Prefer SecretStr inputs over pyright suppression in PostgreSQL config tests.

Passing a plain string forces # pyright: ignore. Using SecretStr keeps type checking clean and aligns with the model’s declared type. Apply this pattern to all constructor calls in this file.

♻️ Proposed fix (apply similarly to other call sites)

-from pydantic import ValidationError
+from pydantic import SecretStr, ValidationError

-    c = PostgreSQLDatabaseConfiguration(
-        db="db", user="user", password="password"
-    )  # pyright: ignore[reportCallIssue]
+    c = PostgreSQLDatabaseConfiguration(
+        db="db", user="user", password=SecretStr("password")
+    )

Also applies to: 36-41, 64-81, 97-105, 107-115

tests/unit/models/config/test_splunk_configuration.py (1)

20-20: Consolidate pyright suppression into a helper function to reduce noise and centralize type-check handling.

The repeated # pyright: ignore[reportCallIssue] comments across seven test cases add visual clutter. Creating a small make_splunk_config() helper moves the suppression to one place while keeping all call sites clean and type-checked.

♻️ Suggested refactor

Add this helper at the top of the test file:

def make_splunk_config(**overrides: Any) -> SplunkConfiguration:
    """Create a SplunkConfiguration instance for testing."""
    return SplunkConfiguration(**overrides)  # pyright: ignore[reportCallIssue]

Then replace all calls:

-cfg = SplunkConfiguration()  # pyright: ignore[reportCallIssue]
+cfg = make_splunk_config()

-cfg = SplunkConfiguration(enabled=False)  # pyright: ignore[reportCallIssue]
+cfg = make_splunk_config(enabled=False)

-SplunkConfiguration(
+make_splunk_config(
     enabled=True,
     url=url,
     token_path=token_file if has_token else None,
     index=index,
-)  # pyright: ignore[reportCallIssue]
+)

And similarly for lines 94–96, 102–104, and 110–112.