fix: Improve knowledge context formatting for better LLM effectiveness

- Make knowledge context more explicit and instructive for LLM
- Add clear instructions to use provided context
- Add comprehensive test coverage for knowledge utils

Fixes #2269

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

tests/knowledge/test_knowledge_utils.py

@@ -0,0 +1,39 @@
+"""Test knowledge utils functionality."""
+
+from typing import Dict, List, Any
+
+import pytest
+
+from crewai.knowledge.utils.knowledge_utils import extract_knowledge_context
+
+
+def test_extract_knowledge_context_with_valid_snippets():
+    """Test extracting knowledge context with valid snippets."""
+    snippets = [
+        {"context": "Fact 1: The sky is blue", "score": 0.9},
+        {"context": "Fact 2: Water is wet", "score": 0.8},
+    ]
+    result = extract_knowledge_context(snippets)
+    expected = "Additional Information: Fact 1: The sky is blue\nFact 2: Water is wet"
+    assert result == expected
+
+
+def test_extract_knowledge_context_with_empty_snippets():
+    """Test extracting knowledge context with empty snippets."""
+    snippets: List[Dict[str, Any]] = []
+    result = extract_knowledge_context(snippets)
+    assert result == ""
+
+
+def test_extract_knowledge_context_with_none_snippets():
+    """Test extracting knowledge context with None snippets."""
+    snippets = [None, {"context": "Valid context"}]  # type: ignore
+    result = extract_knowledge_context(snippets)
+    assert result == "Additional Information: Valid context"
+
+
+def test_extract_knowledge_context_with_missing_context():
+    """Test extracting knowledge context with missing context."""
+    snippets = [{"score": 0.9}, {"context": "Valid context"}]
+    result = extract_knowledge_context(snippets)
+    assert result == "Additional Information: Valid context"