fix lint

Author: zhongyu09

README.md

@@ -825,6 +825,18 @@ uv run pytest test/test_generate_sql.py
 uv run pytest test/test_agent_graph.py
 ```
 
+#### RunLedger Replay Evals
+
+RunLedger provides deterministic replay coverage for the agent/tool protocol
+using recorded JSONL cassettes. It does not require live LLM or warehouse access
+in replay mode.
+
+```bash
+uv run --extra eval runledger run evals/runledger --mode replay --baseline baselines/runledger-openchatbi.json
+```
+
+Run artifacts are written under `runledger_out/`.
+
 ### Pre-commit Hooks
 
 Install pre-commit hooks for automatic code quality checks:

evals/runledger/agent/agent.py

@@ -2,8 +2,7 @@
 import json
 import sys
 from itertools import count
-from typing import Literal
-from typing import Any
+from typing import Any, Literal
 from unittest.mock import MagicMock
 
 from langchain_core.messages import AIMessage, HumanMessage, ToolMessage

evals/runledger/suite.yaml

@@ -20,4 +20,7 @@ budgets:
   max_tool_calls: 8
   max_tool_errors: 0
 
+regression:
+  min_pass_rate: 0.95
+
 baseline_path: ../../baselines/runledger-openchatbi.json

openchatbi/streaming.py

@@ -142,7 +142,7 @@ def _message_list(value: object) -> list[object]:
     if value is None:
         return []
     if value.__class__.__name__ == "Overwrite" and hasattr(value, "value"):
-        value = getattr(value, "value")
+        value = value.value
     if isinstance(value, list | tuple):
         return list(value)
     return [value]

tests/test_tools_search_knowledge.py

@@ -4,6 +4,7 @@
 from unittest.mock import Mock, patch
 
 import pytest
+from pydantic import ValidationError
 
 from openchatbi.tool import search_knowledge as sk
 from openchatbi.tool.search_knowledge import search_knowledge, search_schema, show_schema
@@ -127,7 +128,7 @@ def test_search_knowledge_business_without_dedicated_store(self):
         assert "No dedicated business knowledge available" in result["warnings"][0]
 
     def test_search_knowledge_rejects_invalid_knowledge_base(self):
-        with pytest.raises(Exception):
+        with pytest.raises(ValidationError):
             search_knowledge.invoke(
                 {
                     "reasoning": "invalid kb",