fix: use .get() for all state access to prevent KeyError in graph nodes

Makefile

@@ -14,4 +14,4 @@ build:
 	docker build -t ai-interviewmentor .
 
 docker-run:
-	docker run --env-file .env -p 7860:7860 ai-interviewmentor
+	docker run --name ai-interviewmentor --env-file .env -p 7860:7860 --rm ai-interviewmentor

backend/graph/nodes.py

@@ -32,9 +32,9 @@ MAX_TURNS = 5
 
 
 async def ask_question(state: InterviewState) -> dict:
-    remaining = list(state["questions_remaining"])
-    log.info("ask_question: turn=%d/%d, remaining=%d", state["turn_count"], MAX_TURNS, len(remaining))
-    if not remaining or state["turn_count"] >= MAX_TURNS:
+    remaining = list(state.get("questions_remaining", []))
+    log.info("ask_question: turn=%d/%d, remaining=%d", state.get("turn_count", 0), MAX_TURNS, len(remaining))
+    if not remaining or state.get("turn_count", 0) >= MAX_TURNS:
         log.info("ask_question: skipping (limit reached or no questions)")
         return {"messages": []}
 
@@ -50,9 +50,9 @@ async def ask_question(state: InterviewState) -> dict:
 
     return {
         "questions_remaining": remaining,
-        "questions_asked": state["questions_asked"] + [question["question_text"]],
+        "questions_asked": state.get("questions_asked", []) + [question["question_text"]],
         "messages": [{"role": "assistant", "content": response}],
-        "turn_count": state["turn_count"] + 1,
+        "turn_count": state.get("turn_count", 0) + 1,
         "awaiting_counter_response": False,
     }
 
@@ -126,14 +126,14 @@ async def summarize(state: InterviewState) -> dict:
 
 
 async def generate_report(state: InterviewState) -> dict:
-    log.info("generate_report: questions_asked=%d, turn_count=%d", len(state["questions_asked"]), state["turn_count"])
+    log.info("generate_report: questions_asked=%d, turn_count=%d", len(state.get("questions_asked", [])), state.get("turn_count", 0))
     prompt = build_report_prompt(
-        state["topic_name"],
-        state["questions_asked"],
-        state["student_weak_areas"],
-        state["conversation_summary"],
-        state["past_best_score"],
-        messages=state["messages"],
+        state.get("topic_name", ""),
+        state.get("questions_asked", []),
+        state.get("student_weak_areas", []),
+        state.get("conversation_summary", ""),
+        state.get("past_best_score"),
+        messages=state.get("messages", []),
     )
     raw = await call_llm(prompt, max_tokens=400)
 
@@ -164,9 +164,9 @@ async def generate_report(state: InterviewState) -> dict:
 def route_after_evaluation(state: InterviewState) -> str:
     """Routing function for conditional edges after evaluate_answer."""
     verdict = state.get("last_verdict", "wrong")
-    turn_count = state["turn_count"]
-    questions_remaining = state["questions_remaining"]
-    awaiting_counter = state["awaiting_counter_response"]
+    turn_count = state.get("turn_count", 0)
+    questions_remaining = state.get("questions_remaining", [])
+    awaiting_counter = state.get("awaiting_counter_response", False)
 
     decision = "unknown"