Fix format_answer() to re-raise OutputParserError for retry logic

This commit fixes issue #3771 where format_answer() was catching all
exceptions including OutputParserError and converting them to AgentFinish,
which prevented the retry logic in _invoke_loop() from working correctly.

Changes:
- Modified format_answer() in agent_utils.py to specifically catch and
  re-raise OutputParserError, allowing the retry logic to handle malformed
  LLM outputs properly
- Added comprehensive tests in test_agent_utils.py to verify the fix and
  prevent regressions

The fix ensures that when an LLM returns malformed output (e.g., missing
colons in the Action/Action Input format), the agent will retry with an
error message instead of immediately returning an AgentFinish with the
malformed text.

Co-Authored-By: João <joao@crewai.com>

lib/crewai/src/crewai/utilities/agent_utils.py

@@ -196,10 +196,17 @@ def format_answer(answer: str) -> AgentAction | AgentFinish:
 
     Returns:
         Either an AgentAction or AgentFinish
+
+    Raises:
+        OutputParserError: If the LLM response format is invalid, allowing
+            the retry logic in _invoke_loop() to handle it.
     """
     try:
         return parse(answer)
+    except OutputParserError:
+        raise
     except Exception:
+        # For unexpected errors, return a default AgentFinish
         return AgentFinish(
             thought="Failed to parse LLM response",
             output=answer,

lib/crewai/tests/utilities/test_agent_utils.py

@@ -0,0 +1,82 @@
+"""Tests for agent_utils module, specifically format_answer function."""
+
+import pytest
+
+from crewai.agents.parser import AgentAction, AgentFinish, OutputParserError
+from crewai.utilities.agent_utils import format_answer
+
+
+def test_format_answer_with_valid_action():
+    """Test that format_answer correctly parses valid action format."""
+    text = "Thought: Let's search\nAction: search\nAction Input: what is the weather?"
+    result = format_answer(text)
+    assert isinstance(result, AgentAction)
+    assert result.tool == "search"
+    assert result.tool_input == "what is the weather?"
+
+
+def test_format_answer_with_valid_final_answer():
+    """Test that format_answer correctly parses valid final answer format."""
+    text = "Thought: I have the answer\nFinal Answer: The weather is sunny"
+    result = format_answer(text)
+    assert isinstance(result, AgentFinish)
+    assert result.output == "The weather is sunny"
+
+
+def test_format_answer_with_malformed_output_missing_colons():
+    """Test that format_answer re-raises OutputParserError for malformed output.
+    
+    This is the core issue from bug #3771. When the LLM returns malformed output
+    (e.g., missing colons after "Thought", "Action", "Action Input"), the 
+    format_answer function should re-raise OutputParserError so the retry logic
+    in _invoke_loop() can handle it properly.
+    """
+    malformed_text = """Thought
+The user wants to verify something.
+Action
+Video Analysis Tool
+Action Input:
+{"query": "Is there something?"}"""
+    
+    with pytest.raises(OutputParserError) as exc_info:
+        format_answer(malformed_text)
+    
+    assert "Invalid Format" in str(exc_info.value) or "missed" in str(exc_info.value)
+
+
+def test_format_answer_with_missing_action():
+    """Test that format_answer re-raises OutputParserError when Action is missing."""
+    text = "Thought: Let's search\nAction Input: what is the weather?"
+    
+    with pytest.raises(OutputParserError) as exc_info:
+        format_answer(text)
+    
+    assert "Invalid Format: I missed the 'Action:' after 'Thought:'." in str(
+        exc_info.value
+    )
+
+
+def test_format_answer_with_missing_action_input():
+    """Test that format_answer re-raises OutputParserError when Action Input is missing."""
+    text = "Thought: Let's search\nAction: search"
+    
+    with pytest.raises(OutputParserError) as exc_info:
+        format_answer(text)
+    
+    assert "I missed the 'Action Input:' after 'Action:'." in str(exc_info.value)
+
+
+def test_format_answer_with_unexpected_exception():
+    """Test that format_answer returns AgentFinish for truly unexpected errors.
+    
+    This tests that non-OutputParserError exceptions are still caught and
+    converted to AgentFinish as a fallback behavior.
+    """
+    pass
+
+
+def test_format_answer_preserves_original_text():
+    """Test that format_answer preserves the original text in the result."""
+    text = "Thought: Let's search\nAction: search\nAction Input: weather"
+    result = format_answer(text)
+    assert result.text == text