fix: Frontend-Backend field name mismatch for workflow questions

Root Cause:
- Frontend serialized as "questions"
- Backend expected "question_augmentations"
- Analysis Nodes WITH questions configured sent empty array to backend
- Questions were never added to LLM prompt

Frontend workflowSerializer.js:
- Serialization: questions → question_augmentations (Backend field name)
- Deserialization: question_augmentations → questions (Frontend data object)
- Backward compatible: Falls back to "questions" for old workflows

Backend workflow_executor.py:
- Removed incorrect load_prompt_questions() function (was a misunderstanding)
- Back to original logic: Only use node.question_augmentations
- Simplified normalization logging

Impact:
- Analysis Node questions are now correctly sent to backend
- Questions augment the base prompt as intended
- LLM receives structured questions
- Decision signals are generated and accessible as placeholders

Example:
- Node configures question with id="q21"
- Signal becomes accessible as {{ node_2.signal_q21 }}
- Can be used in Logic Nodes and End Node templates

Issue: Workflow questions not sent to LLM (field name mismatch)
Version: 0.9p (workflow module)
Part 3: End Node Template Engine - Critical Fix

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

backend/workflow_executor.py

@@ -281,21 +281,13 @@ async def execute_node(
             prompt_template = await load_prompt_template(node.prompt_slug, context)
             logger.debug(f"Node {node.id}: Loaded prompt '{node.prompt_slug}'")
 
-            # 2. Parse question_augmentations (Hybrid Model)
+            # 2. Parse question_augmentations
             questions = []
             if node.question_augmentations:
-                # Node-specific questions (override base prompt questions)
+                # Convert list of dicts to JSONB-like format for parser
                 questions_jsonb = [q.model_dump() if hasattr(q, 'model_dump') else q for q in node.question_augmentations]
                 questions = parse_question_augmentations_from_jsonb(questions_jsonb)
-                logger.debug(f"Node {node.id}: {len(questions)} node-specific questions")
+                logger.debug(f"Node {node.id}: {len(questions)} question augmentations")
-            else:
-                # Fallback: Load questions from base prompt (Hybrid Model)
-                base_questions = await load_prompt_questions(node.prompt_slug)
-                if base_questions:
-                    questions = parse_question_augmentations_from_jsonb(base_questions)
-                    logger.debug(f"Node {node.id}: {len(questions)} questions from base prompt '{node.prompt_slug}'")
-                else:
-                    logger.debug(f"Node {node.id}: No questions (neither node-specific nor base prompt)")
 
             # 3. Augment Prompt
             if questions:
@@ -303,10 +295,8 @@ async def execute_node(
                     base_prompt=prompt_template,
                     questions=questions
                 )
-                logger.debug(f"Node {node.id}: Augmented prompt with {len(questions)} questions")
             else:
                 augmented_prompt = prompt_template
-                logger.debug(f"Node {node.id}: No augmentation (no questions)")
 
             # 4. LLM Call
             logger.debug(f"Node {node.id}: Calling LLM")
@@ -322,17 +312,16 @@ async def execute_node(
             # 6. Normalize Signals
             normalized_signals = []
             if parsed["decision_signals"]:
-                # Hybrid Model: Questions (node-specific or base prompt) override Catalog
+                # Hybrid Model: Node-spezifische Questions überschreiben Catalog
                 node_catalog = catalog.copy()
                 if questions:
                     for q in questions:
                         q_dict = q.model_dump() if hasattr(q, 'model_dump') else q
                         node_catalog[q_dict['type']] = {
                             "answer_spectrum": q_dict['answer_spectrum'],
-                            "normalization_rules": None  # Questions haben keine Synonyme
+                            "normalization_rules": None  # Node-Questions haben keine Synonyme
                         }
-                        source = "node-specific" if node.question_augmentations else "base prompt"
+                        logger.debug(f"Node {node.id}: Override catalog for '{q_dict['type']}' with node-specific spectrum")
-                        logger.debug(f"Node {node.id}: Override catalog for '{q_dict['type']}' with {source} spectrum")
 
                 normalized_signals = normalize_all_signals(
                     decision_signals=parsed["decision_signals"],
@@ -827,50 +816,6 @@ async def load_prompt_template(prompt_slug: str, context: Dict[str, Any]) -> str
     return resolved
 
 
-async def load_prompt_questions(prompt_slug: str) -> List[Dict]:
-    """
-    Lädt Fragen aus einem Basis-Prompt (Hybrid Model - Fallback).
-
-    Wenn ein Analysis Node KEINE node-spezifischen Fragen hat,
-    werden die Fragen aus dem referenzierten Basis-Prompt geladen.
-
-    Args:
-        prompt_slug: Slug des Prompts (z.B. "pipeline_body")
-
-    Returns:
-        Liste von Question-Dicts im format:
-        [
-            {
-                "id": "q1",
-                "type": "relevanz",
-                "question": "Ist eine vertiefte Analyse relevant?",
-                "answer_spectrum": ["ja", "nein", "unklar"]
-            },
-            ...
-        ]
-
-    Beispiel:
-        >>> questions = await load_prompt_questions("pipeline_body")
-        >>> len(questions) > 0
-        True
-    """
-    with get_db() as conn:
-        cur = get_cursor(conn)
-        cur.execute(
-            "SELECT questions FROM ai_prompts WHERE slug = %s AND active = true",
-            (prompt_slug,)
-        )
-        row = cur.fetchone()
-        if not row or not row.get('questions'):
-            return []
-
-        questions = row['questions']
-        # PostgreSQL JSONB wird automatisch zu Python list/dict konvertiert
-        if isinstance(questions, list):
-            return questions
-        else:
-            logger.warning(f"Unexpected questions format for {prompt_slug}: {type(questions)}")
-            return []
 
 
 def aggregate_results(node_states: List[NodeExecutionState]) -> Dict[str, Any]:

frontend/src/utils/workflowSerializer.js

@@ -23,7 +23,7 @@ export function serializeToWorkflowGraph(nodes, edges, metadata = {}) {
     ...(node.type === 'analysis' && {
       prompt_slug: node.data.prompt_slug || null,
       prompt_name: node.data.prompt_name || null,
-      questions: node.data.questions || [],
+      question_augmentations: node.data.questions || [],  // Backend erwartet question_augmentations
       fallback_strategy: node.data.fallback_strategy || 'conservative_skip'
     }),
 
@@ -85,7 +85,7 @@ export function deserializeFromWorkflowGraph(jsonbData) {
       ...(node.type === 'analysis' && {
         prompt_slug: node.prompt_slug || node.prompt_id || null, // Fallback für alte Workflows mit prompt_id
         prompt_name: node.prompt_name || null, // Falls vom Backend mitgeliefert
-        questions: node.questions || [],
+        questions: node.question_augmentations || node.questions || [],  // Backend sendet question_augmentations
         fallback_strategy: node.fallback_strategy || 'conservative_skip'
       }),