Python: [Python][sample] sanitize JSON output in reflection pattern before Pydantic parsing (fixes #4718)

python/samples/03-workflows/agents/workflow_as_agent_reflection_pattern.py

@@ -64,6 +64,29 @@ class ReviewResponse:
     approved: bool
 
 
+def _sanitize_json_text(text: str) -> str:
+    """Sanitize LLM output before JSON parsing.
+
+    Handles common issues with gpt-4.1/gpt-5 structured output:
+    - Markdown code fences (```json ... ```)
+    - Tab characters inside string values
+    - Truncated JSON (missing closing brackets/quotes)
+    """
+    text = text.strip()
+    if text.startswith("```"):
+        text = text.split("\n", 1)[-1] if "\n" in text else text[3:]
+        if text.endswith("```"):
+            text = text[:-3].rstrip()
+    text = text.replace("\t", " ")
+    open_brackets = text.count("{") - text.count("}")
+    if open_brackets > 0:
+        text += "}" * open_brackets
+    open_braces = text.count("[") - text.count("]")
+    if open_braces > 0:
+        text += "]" * open_braces
+    return text
+
+
 class Reviewer(Executor):
     """Executor that reviews agent responses and provides structured feedback."""
 
@@ -107,7 +130,9 @@ class _Response(BaseModel):
         print("Reviewer: Sending review request to LLM...")
         response = await self._chat_client.get_response(messages=messages, options={"response_format": _Response})
 
-        parsed = _Response.model_validate_json(response.messages[-1].text)
+        raw = response.messages[-1].text
+        sanitized = _sanitize_json_text(raw)
+        parsed = _Response.model_validate_json(sanitized)
 
         print(f"Reviewer: Review complete - Approved: {parsed.approved}")
         print(f"Reviewer: Feedback: {parsed.feedback}")

python/samples/03-workflows/declarative/README.md

@@ -64,7 +64,6 @@ actions:
 
 ### Agent Invocation
 - `InvokeAzureAgent` - Call an Azure AI agent
-- `InvokePromptAgent` - Call a local prompt agent
 
 ### Tool Invocation
 - `InvokeFunctionTool` - Call a registered Python function