Merge pull request #922 from langchain-ai/yogesh/add-error-normalization-middleware

feat: add error normalization middleware for tool and model calls [Closes #904]

apps/agent/agent/middleware/__init__.py

@@ -0,0 +1,3 @@
+from .tool_error_handler import ToolErrorMiddleware
+
+__all__ = ["ToolErrorMiddleware"]

apps/agent/agent/middleware/tool_error_handler.py

@@ -0,0 +1,104 @@
+"""Tool error handling middleware.
+
+Wraps all tool calls in try/except so that unhandled exceptions are
+returned as error ToolMessages instead of crashing the agent run.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+from collections.abc import Awaitable, Callable
+
+from langchain.agents.middleware.types import (
+    AgentMiddleware,
+    AgentState,
+)
+from langchain_core.messages import ToolMessage
+from langgraph.prebuilt.tool_node import ToolCallRequest
+from langgraph.types import Command
+
+logger = logging.getLogger(__name__)
+
+
+def _get_name(candidate: object) -> str | None:
+    if not candidate:
+        return None
+    if isinstance(candidate, str):
+        return candidate
+    if isinstance(candidate, dict):
+        name = candidate.get("name")
+    else:
+        name = getattr(candidate, "name", None)
+    return name if isinstance(name, str) and name else None
+
+
+def _extract_tool_name(request: ToolCallRequest | None) -> str | None:
+    if request is None:
+        return None
+    for attr in ("tool_call", "tool_name", "name"):
+        name = _get_name(getattr(request, attr, None))
+        if name:
+            return name
+    return None
+
+
+def _to_error_payload(e: Exception, request: ToolCallRequest | None = None) -> dict[str, str]:
+    data: dict[str, str] = {
+        "error": str(e),
+        "error_type": e.__class__.__name__,
+        "status": "error",
+    }
+    tool_name = _extract_tool_name(request)
+    if tool_name:
+        data["name"] = tool_name
+    return data
+
+
+def _get_tool_call_id(request: ToolCallRequest) -> str | None:
+    if isinstance(request.tool_call, dict):
+        return request.tool_call.get("id")
+    return None
+
+
+class ToolErrorMiddleware(AgentMiddleware):
+    """Normalize tool execution errors into predictable payloads.
+
+    Catches any exception thrown during a tool call and converts it into
+    a ToolMessage with status="error" so the LLM can see the failure and
+    self-correct, rather than crashing the entire agent run.
+    """
+
+    state_schema = AgentState
+
+    def wrap_tool_call(
+        self,
+        request: ToolCallRequest,
+        handler: Callable[[ToolCallRequest], ToolMessage | Command],
+    ) -> ToolMessage | Command:
+        try:
+            return handler(request)
+        except Exception as e: 
+            logger.exception("Error during tool call handling; request=%r", request)
+            data = _to_error_payload(e, request)
+            return ToolMessage(
+                content=json.dumps(data),
+                tool_call_id=_get_tool_call_id(request),
+                status="error",
+            )
+
+    async def awrap_tool_call(
+        self,
+        request: ToolCallRequest,
+        handler: Callable[[ToolCallRequest], Awaitable[ToolMessage | Command]],
+    ) -> ToolMessage | Command:
+        try:
+            return await handler(request)
+        except Exception as e:  
+            logger.exception("Error during tool call handling; request=%r", request)
+            data = _to_error_payload(e, request)
+            return ToolMessage(
+                content=json.dumps(data),
+                tool_call_id=_get_tool_call_id(request),
+                status="error",
+            )

apps/agent/agent/server.py

@@ -26,13 +26,14 @@ warnings.filterwarnings("ignore", message=".*Pydantic V1.*", category=UserWarnin
 
 # Now safe to import agent (which imports LangChain modules)
 from deepagents import create_deep_agent
-from .protocol import SandboxBackendProtocol
 
 # Local import for encryption
 from langchain_anthropic import ChatAnthropic
 
 from .encryption import decrypt_token
+from .middleware import ToolErrorMiddleware
 from .prompt import construct_system_prompt
+from .protocol import SandboxBackendProtocol
 from .tools import commit_and_open_pr, fetch_url, http_request
 
 
@@ -959,6 +960,7 @@ async def get_agent(config: RunnableConfig) -> Pregel:  # noqa: PLR0915
         tools=[http_request, fetch_url, commit_and_open_pr],
         backend=sandbox_backend,
         middleware=[
+            ToolErrorMiddleware(),
             check_message_queue_before_model,
             post_to_linear_after_model,
             open_pr_if_needed,

apps/agent/agent/webapp.py

@@ -71,7 +71,7 @@ def get_service_jwt_token_for_user(
 
 LINEAR_TEAM_TO_REPO: dict[str, dict[str, str]] = {
     "Brace's test workspace": {"owner": "langchain-ai", "name": "open-swe"},
-    "Yogesh-dev": {"owner": "aran-yogesh", "name": "nimedge"},
+    "Yogesh-dev": {"owner": "aran-yogesh", "name": "TalkBack"},
 }