rootcause and just throw

src/llama_stack/core/server/runtime_error_sanitizer.py

@@ -1,68 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-
-import re
-from collections.abc import Iterable
-from dataclasses import dataclass
-
-from llama_stack.log import get_logger
-
-logger = get_logger(name=__name__)
-
-
-@dataclass(frozen=True)
-class RuntimeErrorRule:
-    code: str
-    default_message: str
-    regex: re.Pattern[str] | None = None
-    template: str | None = None
-
-    def evaluate(self, error_msg: str) -> str | None:
-        """Return the sanitized message if this rule matches, otherwise None."""
-        if self.regex and (match := self.regex.search(error_msg)):
-            if self.template:
-                try:
-                    return self.template.format(**match.groupdict())
-                except Exception:  # pragma: no cover - defensive
-                    logger.debug("Failed to format sanitized runtime error message", exc_info=True)
-            return self.default_message
-
-        return None
-
-
-@dataclass(frozen=True)
-class SanitizedRuntimeError:
-    code: str
-    message: str
-
-
-MODEL_NOT_FOUND_REGEX = re.compile(r"model ['\"]?(?P<model>[^'\" ]+)['\"]? not found", re.IGNORECASE)
-
-
-RUNTIME_ERROR_RULES: tuple[RuntimeErrorRule, ...] = (
-    RuntimeErrorRule(
-        code="MODEL_NOT_FOUND",
-        default_message="Requested model is unavailable.",
-        regex=MODEL_NOT_FOUND_REGEX,
-        template="Requested model '{model}' is unavailable.",
-    ),
-)
-
-
-def sanitize_runtime_error(
-    error: RuntimeError, rules: Iterable[RuntimeErrorRule] = RUNTIME_ERROR_RULES
-) -> SanitizedRuntimeError | None:
-    """
-    Map internal RuntimeError messages to stable, user-safe error codes/messages.
-    """
-    message = str(error)
-
-    for rule in rules:
-        sanitized_message = rule.evaluate(message)
-        if sanitized_message:
-            return SanitizedRuntimeError(code=rule.code, message=sanitized_message)
-
-    return None

src/llama_stack/core/server/server.py

@@ -45,7 +45,6 @@ from llama_stack.core.request_headers import (
     user_from_scope,
 )
 from llama_stack.core.server.routes import get_all_api_routes
-from llama_stack.core.server.runtime_error_sanitizer import sanitize_runtime_error
 from llama_stack.core.stack import (
     Stack,
     cast_image_name_to_string,
@@ -138,16 +137,9 @@ def translate_exception(exc: Exception) -> HTTPException | RequestValidationErro
         detail = str(exc)
         return HTTPException(status_code=status_code, detail=detail)
     else:
-        detail = "Internal server error: An unexpected error occurred."
-        if isinstance(exc, RuntimeError):
-            sanitized = sanitize_runtime_error(exc)
-            if sanitized:
-                logger.warning("RuntimeError sanitized as %s: %s", sanitized.code, str(exc))
-                detail = f"{sanitized.code}: {sanitized.message}"
-
         return HTTPException(
             status_code=httpx.codes.INTERNAL_SERVER_ERROR,
-            detail=detail,
+            detail="Internal server error: An unexpected error occurred.",
         )
 
 

src/llama_stack/providers/inline/agents/meta_reference/responses/openai_responses.py

@@ -363,7 +363,6 @@ class OpenAIResponsesImpl:
 
         # Structured outputs
         response_format = await convert_response_text_to_chat_response_format(text)
-
         ctx = ChatCompletionContext(
             model=model,
             messages=messages,

src/llama_stack/providers/inline/agents/meta_reference/responses/streaming.py

@@ -16,6 +16,7 @@ from llama_stack_api import (
     ApprovalFilter,
     Inference,
     MCPListToolsTool,
+    ModelNotFoundError,
     OpenAIAssistantMessageParam,
     OpenAIChatCompletion,
     OpenAIChatCompletionChunk,
@@ -323,6 +324,8 @@ class StreamingResponseOrchestrator:
             if last_completion_result and last_completion_result.finish_reason == "length":
                 final_status = "incomplete"
 
+        except ModelNotFoundError:
+            raise
         except Exception as exc:  # noqa: BLE001
             self.final_messages = messages.copy()
             self.sequence_number += 1

tests/unit/server/test_runtime_error_sanitizer.py

@@ -1,24 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-
-from llama_stack.core.server.runtime_error_sanitizer import sanitize_runtime_error
-
-
-def test_model_not_found_is_sanitized():
-    err = RuntimeError("OpenAI response failed: Model 'claude-sonnet-4-5-20250929' not found.")
-
-    sanitized = sanitize_runtime_error(err)
-
-    assert sanitized.code == "MODEL_NOT_FOUND"
-    assert sanitized.message == "Requested model 'claude-sonnet-4-5-20250929' is unavailable."
-
-
-def test_unmapped_runtime_error_defaults_to_internal_error():
-    err = RuntimeError("Unexpected failure in obscure subsystem")
-
-    sanitized = sanitize_runtime_error(err)
-
-    assert sanitized is None