fix: include user question in chunk extraction prompts

scrapegraphai/nodes/generate_answer_node_k_level.py

@@ -149,17 +149,18 @@ def execute(self, state: dict) -> dict:
         ):
             prompt = PromptTemplate(
                 template=template_chunks_prompt,
-                input_variables=["format_instructions"],
+                input_variables=["question"],
                 partial_variables={
                     "content": chunk.get("document"),
                     "chunk_id": i + 1,
+                    "format_instructions": format_instructions,
                 },
             )
             chain_name = f"chunk{i + 1}"
             chains_dict[chain_name] = prompt | self.llm_model
 
         async_runner = RunnableParallel(**chains_dict)
-        batch_results = async_runner.invoke({"format_instructions": user_prompt})
+        batch_results = async_runner.invoke({"question": user_prompt})
 
         merge_prompt = PromptTemplate(
             template=template_merge_prompt,

scrapegraphai/prompts/generate_answer_node_prompts.py

@@ -13,6 +13,7 @@
 and things that will invalidate the dictionary. \n
 Do not start the response with ```json because it will invalidate the postprocessing. \n
 OUTPUT INSTRUCTIONS: {format_instructions}\n
+USER QUESTION: {question}
 Content of {chunk_id}: {content}. \n
 """
 
@@ -56,6 +57,7 @@
 and things that will invalidate the dictionary. \n
 Do not start the response with ```json because it will invalidate the postprocessing. \n
 OUTPUT INSTRUCTIONS: {format_instructions}\n
+USER QUESTION: {question}
 Content of {chunk_id}: {content}. \n
 """
 

tests/test_chunk_prompt_question.py

@@ -0,0 +1,67 @@
+import pytest
+from langchain_core.prompts import PromptTemplate
+from scrapegraphai import prompts
+
+
+@pytest.mark.parametrize(
+    "template_name",
+    ["TEMPLATE_CHUNKS_MD", "TEMPLATE_CHUNKS"],
+)
+def test_chunk_prompts_include_user_question(template_name):
+    """Ensure each chunk receives the user's extraction question."""
+    from langchain_core.prompts import PromptTemplate
+    from scrapegraphai import prompts
+
+    template = getattr(prompts, template_name)
+
+    prompt = PromptTemplate(
+        template=template,
+        input_variables=["question"],
+        partial_variables={
+            "content": "Django and Flask are Python web frameworks.",
+            "chunk_id": 1,
+            "format_instructions": "Return valid JSON.",
+        },
+    )
+
+    rendered = prompt.format(question="Which Python frameworks are mentioned?")
+
+    assert "USER QUESTION: Which Python frameworks are mentioned?" in rendered
+    assert "Django and Flask" in rendered
+
+
+def test_k_level_chunk_receives_question(monkeypatch):
+    """Verify K-level chunk processing includes the question and JSON instructions."""
+    from langchain_core.runnables import RunnableLambda
+    from scrapegraphai.nodes.generate_answer_node_k_level import GenerateAnswerNodeKLevel
+
+    captured_prompts = []
+
+    def fake_llm(prompt):
+        captured_prompts.append(prompt.to_string())
+        return '{"content": "Django"}'
+
+    class FakeResult:
+        id = 1
+        score = 0.9
+
+    class FakeDB:
+        def query(self, collection_name, query_text):
+            return [FakeResult()]
+
+    node = GenerateAnswerNodeKLevel(
+        "user_prompt & docs & vectorial_db",
+        ["answer"],
+        node_config={"llm_model": RunnableLambda(fake_llm), "verbose": False},
+    )
+
+    node.execute({
+        "user_prompt": "Which Python frameworks are mentioned?",
+        "docs": [{"document": "Django is a Python framework."}],
+        "vectorial_db": FakeDB(),
+    })
+
+    assert captured_prompts
+    assert "USER QUESTION: Which Python frameworks are mentioned?" in captured_prompts[0]
+    assert "OUTPUT INSTRUCTIONS:" in captured_prompts[0]
+    assert "OUTPUT INSTRUCTIONS: Return a JSON object." in captured_prompts[0]