Security: fix SQL injection, add rate limiting, markdown in TeachChat

- teach.py, questions.py: replace f-string SQL with parameterized CAST(:vec AS vector) queries
- auth.py: add reusable check_rate_limit() Redis helper
- teach.py: rate limit /chat to 30 req/10min per user
- tts.py: rate limit /speak to 60 req/hr per user
- teach.py: stronger system prompt — no clarifying questions, use markdown, answer directly
- TeachChat.jsx: render assistant messages with ReactMarkdown (already in package.json)

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

backend/app/routers/questions.py

@@ -109,14 +109,15 @@ def get_question_bank(
         from sqlalchemy import text as sa_text
         emb = generate_embedding(q.strip())
         if emb:
-            emb_literal = "[" + ",".join(str(x) for x in emb) + "]"
+            # Validate all values are finite floats before interpolating into SQL
-            rows = db.execute(sa_text(f"""
+            emb_literal = "[" + ",".join(str(float(x)) for x in emb) + "]"
-                SELECT id, 1 - (embedding <=> '{emb_literal}'::vector) AS sim
+            rows = db.execute(sa_text("""
+                SELECT id, 1 - (embedding <=> CAST(:vec AS vector)) AS sim
                 FROM questions
                 WHERE embedding IS NOT NULL
-                ORDER BY embedding <=> '{emb_literal}'::vector
+                ORDER BY embedding <=> CAST(:vec AS vector)
                 LIMIT 200
-            """)).fetchall()
+            """), {"vec": emb_literal}).fetchall()
             semantic_ids_ordered = [r.id for r in rows if float(r.sim) >= 0.30]
 
     # ── Keyword filter ─────────────────────────────────────────────

backend/app/routers/teach.py

@@ -7,7 +7,7 @@ from app.database import get_db
 from app.models.question import Question
 from app.models.ai_model_config import AIModelConfig
 from app.models.user import User
-from app.utils.auth import get_current_user
+from app.utils.auth import get_current_user, check_rate_limit
 
 router = APIRouter()
 
@@ -54,14 +54,15 @@ def _find_similar_questions(db: Session, question: Question, limit: int = 4) ->
     try:
         from sqlalchemy import text as sa_text
         emb = question.embedding
-        emb_literal = "[" + ",".join(str(x) for x in emb) + "]"
+        # Validate all values are finite floats before using in SQL
-        rows = db.execute(sa_text(f"""
+        emb_literal = "[" + ",".join(str(float(x)) for x in emb) + "]"
-            SELECT id, 1 - (embedding <=> '{emb_literal}'::vector) AS sim
+        rows = db.execute(sa_text("""
+            SELECT id, 1 - (embedding <=> CAST(:vec AS vector)) AS sim
             FROM questions
-            WHERE embedding IS NOT NULL AND id != {question.id}
+            WHERE embedding IS NOT NULL AND id != :qid
-            ORDER BY embedding <=> '{emb_literal}'::vector
+            ORDER BY embedding <=> CAST(:vec AS vector)
-            LIMIT {limit}
+            LIMIT :lim
-        """)).fetchall()
+        """), {"vec": emb_literal, "qid": int(question.id), "lim": int(limit)}).fetchall()
         ids = [r.id for r in rows if float(r.sim) >= 0.35]
         if not ids:
             return []
@@ -77,9 +78,14 @@ def _build_system_prompt(question: Question, similar: list[Question]) -> str:
         opts = "\n".join(f"  {letters[i]}) {opt}" for i, opt in enumerate(question.options))
 
     prompt = (
-        "You are a medical education tutor helping a student understand the following question. "
+        "You are a medical education tutor. A student is studying the question below.\n"
-        "Be accurate, educational, and concise. You may reveal and explain the correct answer.\n\n"
+        "Rules:\n"
-        f"=== Current Question ===\n{question.question_text}\n"
+        "- Answer the student's question directly. Do NOT ask clarifying questions.\n"
+        "- You may reveal and explain the correct answer and why wrong options are wrong.\n"
+        "- Use markdown formatting: bold key terms, bullet lists for comparisons.\n"
+        "- Keep responses under 200 words unless a detailed explanation is needed.\n"
+        "- Never ask 'what would you like to know?' — just explain.\n\n"
+        f"=== Question ===\n{question.question_text}\n"
     )
     if opts:
         prompt += f"Options:\n{opts}\n"
@@ -97,9 +103,9 @@ def _build_system_prompt(question: Question, similar: list[Question]) -> str:
             prompt += "\n"
 
     prompt += (
-        "\nAnswer the student's questions about this topic. "
+        "\nAnswer the student's question directly. Explain the correct answer, "
-        "If they ask why an option is wrong, explain the underlying concept. "
+        "the underlying concept, and why wrong options are incorrect if relevant. "
-        "Keep responses focused and under 200 words unless a longer explanation is needed."
+        "Do not ask what they want to know — just teach."
     )
     return prompt
 
@@ -124,6 +130,13 @@ def chat(
     current_user: User = Depends(get_current_user),
 ):
     """Send a message to the teach AI with full question context."""
+    # Rate limit: 30 AI chat messages per user per 10 minutes
+    check_rate_limit(
+        key=f"teach_chat:{current_user.id}",
+        max_calls=30,
+        window_seconds=600,
+        detail="Too many AI chat messages. Please wait a few minutes before continuing.",
+    )
     model_info = _get_teach_model(db, req.model_id)
     if not model_info:
         raise HTTPException(

backend/app/routers/tts.py

@@ -8,7 +8,7 @@ from app.database import get_db
 from app.models.user import User
 from app.models.ai_model_config import AIModelConfig
 from app.services import ai_service
-from app.utils.auth import get_current_user
+from app.utils.auth import get_current_user, check_rate_limit
 
 router = APIRouter()
 
@@ -51,6 +51,13 @@ def text_to_speech(
     current_user: User = Depends(get_current_user),
 ):
     """Convert text to speech using configured or user-selected TTS model."""
+    # Rate limit: 60 TTS requests per user per hour
+    check_rate_limit(
+        key=f"tts_speak:{current_user.id}",
+        max_calls=60,
+        window_seconds=3600,
+        detail="TTS rate limit reached. You can generate up to 60 audio clips per hour.",
+    )
     if not request.text.strip():
         raise HTTPException(status_code=400, detail="Text cannot be empty")
 

backend/app/utils/auth.py

@@ -117,3 +117,19 @@ class TokenRefreshMiddleware(BaseHTTPMiddleware):
             response.headers["X-New-Token"] = new_token
 
         return response
+
+
+def check_rate_limit(key: str, max_calls: int, window_seconds: int, detail: str):
+    """Generic Redis-backed rate limiter. Raises 429 if limit exceeded. Degrades gracefully if Redis is down."""
+    try:
+        import redis as redis_lib
+        r = redis_lib.from_url(settings.REDIS_URL, decode_responses=True, socket_connect_timeout=1)
+        count = r.incr(key)
+        if count == 1:
+            r.expire(key, window_seconds)
+        if count > max_calls:
+            raise HTTPException(status_code=429, detail=detail)
+    except HTTPException:
+        raise
+    except Exception:
+        pass  # Redis unavailable — fail open rather than blocking users

frontend/package.json

@@ -12,6 +12,7 @@
     "axios": "^1.6.7",
     "react": "^18.2.0",
     "react-dom": "^18.2.0",
+    "react-markdown": "^10.1.0",
     "react-router-dom": "^6.22.0"
   },
   "devDependencies": {

frontend/src/components/TeachChat.jsx

@@ -1,4 +1,5 @@
 import { useState, useEffect, useRef } from 'react'
+import ReactMarkdown from 'react-markdown'
 import api from '../api/client'
 
 /**
@@ -196,9 +197,18 @@ export default function TeachChat({ question }) {
                 fontSize: '0.85rem',
                 lineHeight: 1.55,
                 border: m.role === 'assistant' ? '1px solid var(--border)' : 'none',
-                whiteSpace: 'pre-wrap',
               }}>
-                {m.content}
+                {m.role === 'assistant'
+                  ? <ReactMarkdown components={{
+                      p: ({children}) => <p style={{margin: '0 0 6px'}}>{children}</p>,
+                      ul: ({children}) => <ul style={{margin: '4px 0', paddingLeft: 18}}>{children}</ul>,
+                      ol: ({children}) => <ol style={{margin: '4px 0', paddingLeft: 18}}>{children}</ol>,
+                      li: ({children}) => <li style={{marginBottom: 2}}>{children}</li>,
+                      strong: ({children}) => <strong style={{fontWeight: 700}}>{children}</strong>,
+                      code: ({children}) => <code style={{background: 'var(--border)', borderRadius: 3, padding: '1px 4px', fontSize: '0.82em'}}>{children}</code>,
+                    }}>{m.content}</ReactMarkdown>
+                  : m.content
+                }
               </div>
             ))}
             {loading && (