Multi-turn conversation, stop generation, SSE fix, and UI improvements

- Multi-turn context: all council stages now receive conversation history
  (user messages + Stage 3 chairman responses) for coherent follow-ups
- Stop generation: abort streaming mid-request, recover query to input box
- SSE parsing: buffer-based chunking to prevent JSON split across packets
- Atomic storage: user + assistant messages saved together after completion,
  preventing dangling messages on abort
- GFM markdown: tables, strikethrough via remark-gfm plugin + table styles
- Performance: memo user messages and completed assistant messages, only
  re-render the active streaming message
- Model config: gpt-5.2, claude-opus-4.6 as chairman
- Always show input box for multi-turn conversations

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

13 files changed, 594 insertions, 117 deletions

diff --git a/backend/config.py b/backend/config.py
index a9cf7c4..cf8fcb4 100644
--- a/backend/config.py
+++ b/backend/config.py
@@ -10,14 +10,14 @@ OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY")
 
 # Council members - list of OpenRouter model identifiers
 COUNCIL_MODELS = [
-    "openai/gpt-5.1",
+    "openai/gpt-5.2",
     "google/gemini-3-pro-preview",
-    "anthropic/claude-sonnet-4.5",
+    "anthropic/claude-opus-4.6",
     "x-ai/grok-4",
 ]
 
 # Chairman model - synthesizes final response
-CHAIRMAN_MODEL = "google/gemini-3-pro-preview"
+CHAIRMAN_MODEL = "anthropic/claude-opus-4.6"
 
 # OpenRouter API endpoint
 OPENROUTER_API_URL = "https://openrouter.ai/api/v1/chat/completions"
diff --git a/backend/council.py b/backend/council.py
index 5069abe..6facbd8 100644
--- a/backend/council.py
+++ b/backend/council.py
@@ -1,21 +1,46 @@
 """3-stage LLM Council orchestration."""
 
-from typing import List, Dict, Any, Tuple
+from typing import List, Dict, Any, Tuple, Optional
 from .openrouter import query_models_parallel, query_model
 from .config import COUNCIL_MODELS, CHAIRMAN_MODEL
 
 
-async def stage1_collect_responses(user_query: str) -> List[Dict[str, Any]]:
+def _build_messages(
+    conversation_history: Optional[List[Dict[str, str]]],
+    current_content: str
+) -> List[Dict[str, str]]:
+    """
+    Build a messages list with conversation history + current user message.
+
+    Args:
+        conversation_history: List of {"role": "user"/"assistant", "content": ...} dicts
+        current_content: The current message content to append as user
+
+    Returns:
+        Messages list for the OpenRouter API
+    """
+    messages = []
+    if conversation_history:
+        messages.extend(conversation_history)
+    messages.append({"role": "user", "content": current_content})
+    return messages
+
+
+async def stage1_collect_responses(
+    user_query: str,
+    conversation_history: Optional[List[Dict[str, str]]] = None
+) -> List[Dict[str, Any]]:
     """
     Stage 1: Collect individual responses from all council models.
 
     Args:
         user_query: The user's question
+        conversation_history: Optional list of prior conversation messages
 
     Returns:
         List of dicts with 'model' and 'response' keys
     """
-    messages = [{"role": "user", "content": user_query}]
+    messages = _build_messages(conversation_history, user_query)
 
     # Query all models in parallel
     responses = await query_models_parallel(COUNCIL_MODELS, messages)
@@ -34,7 +59,8 @@ async def stage1_collect_responses(user_query: str) -> List[Dict[str, Any]]:
 
 async def stage2_collect_rankings(
     user_query: str,
-    stage1_results: List[Dict[str, Any]]
+    stage1_results: List[Dict[str, Any]],
+    conversation_history: Optional[List[Dict[str, str]]] = None
 ) -> Tuple[List[Dict[str, Any]], Dict[str, str]]:
     """
     Stage 2: Each model ranks the anonymized responses.
@@ -42,6 +68,7 @@ async def stage2_collect_rankings(
     Args:
         user_query: The original user query
         stage1_results: Results from Stage 1
+        conversation_history: Optional list of prior conversation messages
 
     Returns:
         Tuple of (rankings list, label_to_model mapping)
@@ -92,7 +119,7 @@ FINAL RANKING:
 
 Now provide your evaluation and ranking:"""
 
-    messages = [{"role": "user", "content": ranking_prompt}]
+    messages = _build_messages(conversation_history, ranking_prompt)
 
     # Get rankings from all council models in parallel
     responses = await query_models_parallel(COUNCIL_MODELS, messages)
@@ -115,7 +142,8 @@ Now provide your evaluation and ranking:"""
 async def stage3_synthesize_final(
     user_query: str,
     stage1_results: List[Dict[str, Any]],
-    stage2_results: List[Dict[str, Any]]
+    stage2_results: List[Dict[str, Any]],
+    conversation_history: Optional[List[Dict[str, str]]] = None
 ) -> Dict[str, Any]:
     """
     Stage 3: Chairman synthesizes final response.
@@ -124,6 +152,7 @@ async def stage3_synthesize_final(
         user_query: The original user query
         stage1_results: Individual model responses from Stage 1
         stage2_results: Rankings from Stage 2
+        conversation_history: Optional list of prior conversation messages
 
     Returns:
         Dict with 'model' and 'response' keys
@@ -156,7 +185,7 @@ Your task as Chairman is to synthesize all of this information into a single, co
 
 Provide a clear, well-reasoned final answer that represents the council's collective wisdom:"""
 
-    messages = [{"role": "user", "content": chairman_prompt}]
+    messages = _build_messages(conversation_history, chairman_prompt)
 
     # Query the chairman model
     response = await query_model(CHAIRMAN_MODEL, messages)
@@ -293,18 +322,22 @@ Title:"""
     return title
 
 
-async def run_full_council(user_query: str) -> Tuple[List, List, Dict, Dict]:
+async def run_full_council(
+    user_query: str,
+    conversation_history: Optional[List[Dict[str, str]]] = None
+) -> Tuple[List, List, Dict, Dict]:
     """
     Run the complete 3-stage council process.
 
     Args:
         user_query: The user's question
+        conversation_history: Optional list of prior conversation messages
 
     Returns:
         Tuple of (stage1_results, stage2_results, stage3_result, metadata)
     """
     # Stage 1: Collect individual responses
-    stage1_results = await stage1_collect_responses(user_query)
+    stage1_results = await stage1_collect_responses(user_query, conversation_history)
 
     # If no models responded successfully, return error
     if not stage1_results:
@@ -314,7 +347,7 @@ async def run_full_council(user_query: str) -> Tuple[List, List, Dict, Dict]:
         }, {}
 
     # Stage 2: Collect rankings
-    stage2_results, label_to_model = await stage2_collect_rankings(user_query, stage1_results)
+    stage2_results, label_to_model = await stage2_collect_rankings(user_query, stage1_results, conversation_history)
 
     # Calculate aggregate rankings
     aggregate_rankings = calculate_aggregate_rankings(stage2_results, label_to_model)
@@ -323,7 +356,8 @@ async def run_full_council(user_query: str) -> Tuple[List, List, Dict, Dict]:
     stage3_result = await stage3_synthesize_final(
         user_query,
         stage1_results,
-        stage2_results
+        stage2_results,
+        conversation_history
     )
 
     # Prepare metadata
diff --git a/backend/main.py b/backend/main.py
index e33ce59..40353dd 100644
--- a/backend/main.py
+++ b/backend/main.py
@@ -14,6 +14,20 @@ from .council import run_full_council, generate_conversation_title, stage1_colle
 
 app = FastAPI(title="LLM Council API")
 
+
+def _extract_conversation_history(conversation: Dict[str, Any]) -> List[Dict[str, str]]:
+    """
+    Extract conversation history as a flat messages list for multi-turn context.
+    User messages use their content; assistant messages use the Stage 3 (chairman) response.
+    """
+    history = []
+    for msg in conversation["messages"]:
+        if msg["role"] == "user":
+            history.append({"role": "user", "content": msg["content"]})
+        elif msg["role"] == "assistant" and msg.get("stage3"):
+            history.append({"role": "assistant", "content": msg["stage3"].get("response", "")})
+    return history
+
 # Enable CORS for local development
 app.add_middleware(
     CORSMiddleware,
@@ -93,20 +107,21 @@ async def send_message(conversation_id: str, request: SendMessageRequest):
     # Check if this is the first message
     is_first_message = len(conversation["messages"]) == 0
 
-    # Add user message
-    storage.add_user_message(conversation_id, request.content)
-
     # If this is the first message, generate a title
     if is_first_message:
         title = await generate_conversation_title(request.content)
         storage.update_conversation_title(conversation_id, title)
 
+    # Build conversation history for multi-turn context
+    conversation_history = _extract_conversation_history(conversation)
+
     # Run the 3-stage council process
     stage1_results, stage2_results, stage3_result, metadata = await run_full_council(
-        request.content
+        request.content, conversation_history
     )
 
-    # Add assistant message with all stages
+    # Save user + assistant messages together only after full completion
+    storage.add_user_message(conversation_id, request.content)
     storage.add_assistant_message(
         conversation_id,
         stage1_results,
@@ -137,11 +152,11 @@ async def send_message_stream(conversation_id: str, request: SendMessageRequest)
     # Check if this is the first message
     is_first_message = len(conversation["messages"]) == 0
 
+    # Build conversation history for multi-turn context
+    conversation_history = _extract_conversation_history(conversation)
+
     async def event_generator():
         try:
-            # Add user message
-            storage.add_user_message(conversation_id, request.content)
-
             # Start title generation in parallel (don't await yet)
             title_task = None
             if is_first_message:
@@ -149,18 +164,18 @@ async def send_message_stream(conversation_id: str, request: SendMessageRequest)
 
             # Stage 1: Collect responses
             yield f"data: {json.dumps({'type': 'stage1_start'})}\n\n"
-            stage1_results = await stage1_collect_responses(request.content)
+            stage1_results = await stage1_collect_responses(request.content, conversation_history)
             yield f"data: {json.dumps({'type': 'stage1_complete', 'data': stage1_results})}\n\n"
 
             # Stage 2: Collect rankings
             yield f"data: {json.dumps({'type': 'stage2_start'})}\n\n"
-            stage2_results, label_to_model = await stage2_collect_rankings(request.content, stage1_results)
+            stage2_results, label_to_model = await stage2_collect_rankings(request.content, stage1_results, conversation_history)
             aggregate_rankings = calculate_aggregate_rankings(stage2_results, label_to_model)
             yield f"data: {json.dumps({'type': 'stage2_complete', 'data': stage2_results, 'metadata': {'label_to_model': label_to_model, 'aggregate_rankings': aggregate_rankings}})}\n\n"
 
             # Stage 3: Synthesize final answer
             yield f"data: {json.dumps({'type': 'stage3_start'})}\n\n"
-            stage3_result = await stage3_synthesize_final(request.content, stage1_results, stage2_results)
+            stage3_result = await stage3_synthesize_final(request.content, stage1_results, stage2_results, conversation_history)
             yield f"data: {json.dumps({'type': 'stage3_complete', 'data': stage3_result})}\n\n"
 
             # Wait for title generation if it was started
@@ -169,7 +184,8 @@ async def send_message_stream(conversation_id: str, request: SendMessageRequest)
                 storage.update_conversation_title(conversation_id, title)
                 yield f"data: {json.dumps({'type': 'title_complete', 'data': {'title': title}})}\n\n"
 
-            # Save complete assistant message
+            # Save user + assistant messages together only after full completion
+            storage.add_user_message(conversation_id, request.content)
             storage.add_assistant_message(
                 conversation_id,
                 stage1_results,
diff --git a/frontend/package-lock.json b/frontend/package-lock.json
index a6a7c34..0f1c4d9 100644
--- a/frontend/package-lock.json
+++ b/frontend/package-lock.json
@@ -10,7 +10,8 @@
       "dependencies": {
         "react": "^19.2.0",
         "react-dom": "^19.2.0",
-        "react-markdown": "^10.1.0"
+        "react-markdown": "^10.1.0",
+        "remark-gfm": "^4.0.1"
       },
       "devDependencies": {
         "@eslint/js": "^9.39.1",
@@ -2391,6 +2392,44 @@
         "yallist": "^3.0.2"
       }
     },
+    "node_modules/markdown-table": {
+      "version": "3.0.4",
+      "resolved": "https://registry.npmjs.org/markdown-table/-/markdown-table-3.0.4.tgz",
+      "integrity": "sha512-wiYz4+JrLyb/DqW2hkFJxP7Vd7JuTDm77fvbM8VfEQdmSMqcImWeeRbHwZjBjIFki/VaMK2BhFi7oUUZeM5bqw==",
+      "license": "MIT",
+      "funding": {
+        "type": "github",
+        "url": "https://github.com/sponsors/wooorm"
+      }
+    },
+    "node_modules/mdast-util-find-and-replace": {
+      "version": "3.0.2",
+      "resolved": "https://registry.npmjs.org/mdast-util-find-and-replace/-/mdast-util-find-and-replace-3.0.2.tgz",
+      "integrity": "sha512-Tmd1Vg/m3Xz43afeNxDIhWRtFZgM2VLyaf4vSTYwudTyeuTneoL3qtWMA5jeLyz/O1vDJmmV4QuScFCA2tBPwg==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/mdast": "^4.0.0",
+        "escape-string-regexp": "^5.0.0",
+        "unist-util-is": "^6.0.0",
+        "unist-util-visit-parents": "^6.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/mdast-util-find-and-replace/node_modules/escape-string-regexp": {
+      "version": "5.0.0",
+      "resolved": "https://registry.npmjs.org/escape-string-regexp/-/escape-string-regexp-5.0.0.tgz",
+      "integrity": "sha512-/veY75JbMK4j1yjvuUxuVsiS/hr/4iHs9FTT6cgTexxdE0Ly/glccBAkloH/DofkjRbZU3bnoj38mOmhkZ0lHw==",
+      "license": "MIT",
+      "engines": {
+        "node": ">=12"
+      },
+      "funding": {
+        "url": "https://github.com/sponsors/sindresorhus"
+      }
+    },
     "node_modules/mdast-util-from-markdown": {
       "version": "2.0.2",
       "resolved": "https://registry.npmjs.org/mdast-util-from-markdown/-/mdast-util-from-markdown-2.0.2.tgz",
@@ -2414,6 +2453,107 @@
         "url": "https://opencollective.com/unified"
       }
     },
+    "node_modules/mdast-util-gfm": {
+      "version": "3.1.0",
+      "resolved": "https://registry.npmjs.org/mdast-util-gfm/-/mdast-util-gfm-3.1.0.tgz",
+      "integrity": "sha512-0ulfdQOM3ysHhCJ1p06l0b0VKlhU0wuQs3thxZQagjcjPrlFRqY215uZGHHJan9GEAXd9MbfPjFJz+qMkVR6zQ==",
+      "license": "MIT",
+      "dependencies": {
+        "mdast-util-from-markdown": "^2.0.0",
+        "mdast-util-gfm-autolink-literal": "^2.0.0",
+        "mdast-util-gfm-footnote": "^2.0.0",
+        "mdast-util-gfm-strikethrough": "^2.0.0",
+        "mdast-util-gfm-table": "^2.0.0",
+        "mdast-util-gfm-task-list-item": "^2.0.0",
+        "mdast-util-to-markdown": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/mdast-util-gfm-autolink-literal": {
+      "version": "2.0.1",
+      "resolved": "https://registry.npmjs.org/mdast-util-gfm-autolink-literal/-/mdast-util-gfm-autolink-literal-2.0.1.tgz",
+      "integrity": "sha512-5HVP2MKaP6L+G6YaxPNjuL0BPrq9orG3TsrZ9YXbA3vDw/ACI4MEsnoDpn6ZNm7GnZgtAcONJyPhOP8tNJQavQ==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/mdast": "^4.0.0",
+        "ccount": "^2.0.0",
+        "devlop": "^1.0.0",
+        "mdast-util-find-and-replace": "^3.0.0",
+        "micromark-util-character": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/mdast-util-gfm-footnote": {
+      "version": "2.1.0",
+      "resolved": "https://registry.npmjs.org/mdast-util-gfm-footnote/-/mdast-util-gfm-footnote-2.1.0.tgz",
+      "integrity": "sha512-sqpDWlsHn7Ac9GNZQMeUzPQSMzR6Wv0WKRNvQRg0KqHh02fpTz69Qc1QSseNX29bhz1ROIyNyxExfawVKTm1GQ==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/mdast": "^4.0.0",
+        "devlop": "^1.1.0",
+        "mdast-util-from-markdown": "^2.0.0",
+        "mdast-util-to-markdown": "^2.0.0",
+        "micromark-util-normalize-identifier": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/mdast-util-gfm-strikethrough": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/mdast-util-gfm-strikethrough/-/mdast-util-gfm-strikethrough-2.0.0.tgz",
+      "integrity": "sha512-mKKb915TF+OC5ptj5bJ7WFRPdYtuHv0yTRxK2tJvi+BDqbkiG7h7u/9SI89nRAYcmap2xHQL9D+QG/6wSrTtXg==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/mdast": "^4.0.0",
+        "mdast-util-from-markdown": "^2.0.0",
+        "mdast-util-to-markdown": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/mdast-util-gfm-table": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/mdast-util-gfm-table/-/mdast-util-gfm-table-2.0.0.tgz",
+      "integrity": "sha512-78UEvebzz/rJIxLvE7ZtDd/vIQ0RHv+3Mh5DR96p7cS7HsBhYIICDBCu8csTNWNO6tBWfqXPWekRuj2FNOGOZg==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/mdast": "^4.0.0",
+        "devlop": "^1.0.0",
+        "markdown-table": "^3.0.0",
+        "mdast-util-from-markdown": "^2.0.0",
+        "mdast-util-to-markdown": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/mdast-util-gfm-task-list-item": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/mdast-util-gfm-task-list-item/-/mdast-util-gfm-task-list-item-2.0.0.tgz",
+      "integrity": "sha512-IrtvNvjxC1o06taBAVJznEnkiHxLFTzgonUdy8hzFVeDun0uTjxxrRGVaNFqkU1wJR3RBPEfsxmU6jDWPofrTQ==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/mdast": "^4.0.0",
+        "devlop": "^1.0.0",
+        "mdast-util-from-markdown": "^2.0.0",
+        "mdast-util-to-markdown": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
     "node_modules/mdast-util-mdx-expression": {
       "version": "2.0.1",
       "resolved": "https://registry.npmjs.org/mdast-util-mdx-expression/-/mdast-util-mdx-expression-2.0.1.tgz",
@@ -2603,6 +2743,127 @@
         "micromark-util-types": "^2.0.0"
       }
     },
+    "node_modules/micromark-extension-gfm": {
+      "version": "3.0.0",
+      "resolved": "https://registry.npmjs.org/micromark-extension-gfm/-/micromark-extension-gfm-3.0.0.tgz",
+      "integrity": "sha512-vsKArQsicm7t0z2GugkCKtZehqUm31oeGBV/KVSorWSy8ZlNAv7ytjFhvaryUiCUJYqs+NoE6AFhpQvBTM6Q4w==",
+      "license": "MIT",
+      "dependencies": {
+        "micromark-extension-gfm-autolink-literal": "^2.0.0",
+        "micromark-extension-gfm-footnote": "^2.0.0",
+        "micromark-extension-gfm-strikethrough": "^2.0.0",
+        "micromark-extension-gfm-table": "^2.0.0",
+        "micromark-extension-gfm-tagfilter": "^2.0.0",
+        "micromark-extension-gfm-task-list-item": "^2.0.0",
+        "micromark-util-combine-extensions": "^2.0.0",
+        "micromark-util-types": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/micromark-extension-gfm-autolink-literal": {
+      "version": "2.1.0",
+      "resolved": "https://registry.npmjs.org/micromark-extension-gfm-autolink-literal/-/micromark-extension-gfm-autolink-literal-2.1.0.tgz",
+      "integrity": "sha512-oOg7knzhicgQ3t4QCjCWgTmfNhvQbDDnJeVu9v81r7NltNCVmhPy1fJRX27pISafdjL+SVc4d3l48Gb6pbRypw==",
+      "license": "MIT",
+      "dependencies": {
+        "micromark-util-character": "^2.0.0",
+        "micromark-util-sanitize-uri": "^2.0.0",
+        "micromark-util-symbol": "^2.0.0",
+        "micromark-util-types": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/micromark-extension-gfm-footnote": {
+      "version": "2.1.0",
+      "resolved": "https://registry.npmjs.org/micromark-extension-gfm-footnote/-/micromark-extension-gfm-footnote-2.1.0.tgz",
+      "integrity": "sha512-/yPhxI1ntnDNsiHtzLKYnE3vf9JZ6cAisqVDauhp4CEHxlb4uoOTxOCJ+9s51bIB8U1N1FJ1RXOKTIlD5B/gqw==",
+      "license": "MIT",
+      "dependencies": {
+        "devlop": "^1.0.0",
+        "micromark-core-commonmark": "^2.0.0",
+        "micromark-factory-space": "^2.0.0",
+        "micromark-util-character": "^2.0.0",
+        "micromark-util-normalize-identifier": "^2.0.0",
+        "micromark-util-sanitize-uri": "^2.0.0",
+        "micromark-util-symbol": "^2.0.0",
+        "micromark-util-types": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/micromark-extension-gfm-strikethrough": {
+      "version": "2.1.0",
+      "resolved": "https://registry.npmjs.org/micromark-extension-gfm-strikethrough/-/micromark-extension-gfm-strikethrough-2.1.0.tgz",
+      "integrity": "sha512-ADVjpOOkjz1hhkZLlBiYA9cR2Anf8F4HqZUO6e5eDcPQd0Txw5fxLzzxnEkSkfnD0wziSGiv7sYhk/ktvbf1uw==",
+      "license": "MIT",
+      "dependencies": {
+        "devlop": "^1.0.0",
+        "micromark-util-chunked": "^2.0.0",
+        "micromark-util-classify-character": "^2.0.0",
+        "micromark-util-resolve-all": "^2.0.0",
+        "micromark-util-symbol": "^2.0.0",
+        "micromark-util-types": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/micromark-extension-gfm-table": {
+      "version": "2.1.1",
+      "resolved": "https://registry.npmjs.org/micromark-extension-gfm-table/-/micromark-extension-gfm-table-2.1.1.tgz",
+      "integrity": "sha512-t2OU/dXXioARrC6yWfJ4hqB7rct14e8f7m0cbI5hUmDyyIlwv5vEtooptH8INkbLzOatzKuVbQmAYcbWoyz6Dg==",
+      "license": "MIT",
+      "dependencies": {
+        "devlop": "^1.0.0",
+        "micromark-factory-space": "^2.0.0",
+        "micromark-util-character": "^2.0.0",
+        "micromark-util-symbol": "^2.0.0",
+        "micromark-util-types": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/micromark-extension-gfm-tagfilter": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/micromark-extension-gfm-tagfilter/-/micromark-extension-gfm-tagfilter-2.0.0.tgz",
+      "integrity": "sha512-xHlTOmuCSotIA8TW1mDIM6X2O1SiX5P9IuDtqGonFhEK0qgRI4yeC6vMxEV2dgyr2TiD+2PQ10o+cOhdVAcwfg==",
+      "license": "MIT",
+      "dependencies": {
+        "micromark-util-types": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
+    "node_modules/micromark-extension-gfm-task-list-item": {
+      "version": "2.1.0",
+      "resolved": "https://registry.npmjs.org/micromark-extension-gfm-task-list-item/-/micromark-extension-gfm-task-list-item-2.1.0.tgz",
+      "integrity": "sha512-qIBZhqxqI6fjLDYFTBIa4eivDMnP+OZqsNwmQ3xNLE4Cxwc+zfQEfbs6tzAo2Hjq+bh6q5F+Z8/cksrLFYWQQw==",
+      "license": "MIT",
+      "dependencies": {
+        "devlop": "^1.0.0",
+        "micromark-factory-space": "^2.0.0",
+        "micromark-util-character": "^2.0.0",
+        "micromark-util-symbol": "^2.0.0",
+        "micromark-util-types": "^2.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
     "node_modules/micromark-factory-destination": {
       "version": "2.0.1",
       "resolved": "https://registry.npmjs.org/micromark-factory-destination/-/micromark-factory-destination-2.0.1.tgz",
@@ -3231,6 +3492,24 @@
         "node": ">=0.10.0"
       }
     },
+    "node_modules/remark-gfm": {
+      "version": "4.0.1",
+      "resolved": "https://registry.npmjs.org/remark-gfm/-/remark-gfm-4.0.1.tgz",
+      "integrity": "sha512-1quofZ2RQ9EWdeN34S79+KExV1764+wCUGop5CPL1WGdD0ocPpu91lzPGbwWMECpEpd42kJGQwzRfyov9j4yNg==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/mdast": "^4.0.0",
+        "mdast-util-gfm": "^3.0.0",
+        "micromark-extension-gfm": "^3.0.0",
+        "remark-parse": "^11.0.0",
+        "remark-stringify": "^11.0.0",
+        "unified": "^11.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
     "node_modules/remark-parse": {
       "version": "11.0.0",
       "resolved": "https://registry.npmjs.org/remark-parse/-/remark-parse-11.0.0.tgz",
@@ -3262,6 +3541,21 @@
         "url": "https://opencollective.com/unified"
       }
     },
+    "node_modules/remark-stringify": {
+      "version": "11.0.0",
+      "resolved": "https://registry.npmjs.org/remark-stringify/-/remark-stringify-11.0.0.tgz",
+      "integrity": "sha512-1OSmLd3awB/t8qdoEOMazZkNsfVTeY4fTsgzcQFdXNq8ToTN4ZGwrMnlda4K6smTFKD+GRV6O48i6Z4iKgPPpw==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/mdast": "^4.0.0",
+        "mdast-util-to-markdown": "^2.0.0",
+        "unified": "^11.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
     "node_modules/resolve-from": {
       "version": "4.0.0",
       "resolved": "https://registry.npmjs.org/resolve-from/-/resolve-from-4.0.0.tgz",
diff --git a/frontend/package.json b/frontend/package.json
index 89fd2c8..1963a90 100644
--- a/frontend/package.json
+++ b/frontend/package.json
@@ -12,7 +12,8 @@
   "dependencies": {
     "react": "^19.2.0",
     "react-dom": "^19.2.0",
-    "react-markdown": "^10.1.0"
+    "react-markdown": "^10.1.0",
+    "remark-gfm": "^4.0.1"
   },
   "devDependencies": {
     "@eslint/js": "^9.39.1",
diff --git a/frontend/src/App.jsx b/frontend/src/App.jsx
index 1954155..8b7a547 100644
--- a/frontend/src/App.jsx
+++ b/frontend/src/App.jsx
@@ -1,4 +1,4 @@
-import { useState, useEffect } from 'react';
+import { useState, useEffect, useRef } from 'react';
 import Sidebar from './components/Sidebar';
 import ChatInterface from './components/ChatInterface';
 import { api } from './api';
@@ -9,6 +9,8 @@ function App() {
   const [currentConversationId, setCurrentConversationId] = useState(null);
   const [currentConversation, setCurrentConversation] = useState(null);
   const [isLoading, setIsLoading] = useState(false);
+  const [pendingInput, setPendingInput] = useState(null);
+  const abortControllerRef = useRef(null);
 
   // Load conversations on mount
   useEffect(() => {
@@ -57,9 +59,36 @@ function App() {
     setCurrentConversationId(id);
   };
 
+  const handleStopGeneration = () => {
+    if (abortControllerRef.current) {
+      abortControllerRef.current.abort();
+      abortControllerRef.current = null;
+
+      // Recover the last user message into the input box and remove the incomplete pair
+      setCurrentConversation((prev) => {
+        const messages = [...prev.messages];
+        // Find the last user message to recover its content
+        let recoveredContent = '';
+        // Remove trailing assistant message (incomplete)
+        if (messages.length > 0 && messages[messages.length - 1].role === 'assistant') {
+          messages.pop();
+        }
+        // Remove the user message and recover its text
+        if (messages.length > 0 && messages[messages.length - 1].role === 'user') {
+          const userMsg = messages.pop();
+          recoveredContent = userMsg.content;
+        }
+        setPendingInput(recoveredContent);
+        return { ...prev, messages };
+      });
+    }
+  };
+
   const handleSendMessage = async (content) => {
     if (!currentConversationId) return;
 
+    const controller = new AbortController();
+    abortControllerRef.current = controller;
     setIsLoading(true);
     try {
       // Optimistically add user message to UI
@@ -91,6 +120,7 @@ function App() {
 
       // Send message with streaming
       await api.sendMessageStream(currentConversationId, content, (eventType, event) => {
+        if (controller.signal.aborted) return;
         switch (eventType) {
           case 'stage1_start':
             setCurrentConversation((prev) => {
@@ -169,15 +199,20 @@ function App() {
           default:
             console.log('Unknown event type:', eventType);
         }
-      });
+      }, controller.signal);
     } catch (error) {
-      console.error('Failed to send message:', error);
-      // Remove optimistic messages on error
-      setCurrentConversation((prev) => ({
-        ...prev,
-        messages: prev.messages.slice(0, -2),
-      }));
+      if (error.name === 'AbortError') {
+        // User stopped generation — handleStopGeneration already cleaned up messages
+      } else {
+        console.error('Failed to send message:', error);
+        // Remove optimistic messages on error
+        setCurrentConversation((prev) => ({
+          ...prev,
+          messages: prev.messages.slice(0, -2),
+        }));
+      }
       setIsLoading(false);
+      abortControllerRef.current = null;
     }
   };
 
@@ -192,7 +227,10 @@ function App() {
       <ChatInterface
         conversation={currentConversation}
         onSendMessage={handleSendMessage}
+        onStopGeneration={handleStopGeneration}
         isLoading={isLoading}
+        pendingInput={pendingInput}
+        onPendingInputConsumed={() => setPendingInput(null)}
       />
     </div>
   );
diff --git a/frontend/src/api.js b/frontend/src/api.js
index 87ec685..a53ed90 100644
--- a/frontend/src/api.js
+++ b/frontend/src/api.js
@@ -73,7 +73,7 @@ export const api = {
    * @param {function} onEvent - Callback function for each event: (eventType, data) => void
    * @returns {Promise<void>}
    */
-  async sendMessageStream(conversationId, content, onEvent) {
+  async sendMessageStream(conversationId, content, onEvent, signal) {
     const response = await fetch(
       `${API_BASE}/api/conversations/${conversationId}/message/stream`,
       {
@@ -82,6 +82,7 @@ export const api = {
           'Content-Type': 'application/json',
         },
         body: JSON.stringify({ content }),
+        signal,
       }
     );
 
@@ -91,22 +92,30 @@ export const api = {
 
     const reader = response.body.getReader();
     const decoder = new TextDecoder();
+    let buffer = '';
 
     while (true) {
       const { done, value } = await reader.read();
       if (done) break;
 
-      const chunk = decoder.decode(value);
-      const lines = chunk.split('\n');
+      buffer += decoder.decode(value, { stream: true });
 
-      for (const line of lines) {
-        if (line.startsWith('data: ')) {
-          const data = line.slice(6);
-          try {
-            const event = JSON.parse(data);
-            onEvent(event.type, event);
-          } catch (e) {
-            console.error('Failed to parse SSE event:', e);
+      // Split on double newline (SSE event boundary)
+      const parts = buffer.split('\n\n');
+      // Last part may be incomplete — keep it in buffer
+      buffer = parts.pop();
+
+      for (const part of parts) {
+        const lines = part.split('\n');
+        for (const line of lines) {
+          if (line.startsWith('data: ')) {
+            const data = line.slice(6);
+            try {
+              const event = JSON.parse(data);
+              onEvent(event.type, event);
+            } catch (e) {
+              console.error('Failed to parse SSE event:', e);
+            }
           }
         }
       }
diff --git a/frontend/src/components/ChatInterface.css b/frontend/src/components/ChatInterface.css
index 0d01300..63bc75d 100644
--- a/frontend/src/components/ChatInterface.css
+++ b/frontend/src/components/ChatInterface.css
@@ -161,3 +161,22 @@
   background: #ccc;
   border-color: #ccc;
 }
+
+.stop-button {
+  padding: 14px 28px;
+  background: #e24a4a;
+  border: 1px solid #e24a4a;
+  border-radius: 8px;
+  color: #fff;
+  font-size: 15px;
+  font-weight: 600;
+  cursor: pointer;
+  transition: background 0.2s;
+  white-space: nowrap;
+  align-self: flex-end;
+}
+
+.stop-button:hover {
+  background: #c93636;
+  border-color: #c93636;
+}
diff --git a/frontend/src/components/ChatInterface.jsx b/frontend/src/components/ChatInterface.jsx
index 3ae796c..5f431c2 100644
--- a/frontend/src/components/ChatInterface.jsx
+++ b/frontend/src/components/ChatInterface.jsx
@@ -1,25 +1,102 @@
-import { useState, useEffect, useRef } from 'react';
+import { useState, useEffect, useRef, memo } from 'react';
 import ReactMarkdown from 'react-markdown';
+import remarkGfm from 'remark-gfm';
 import Stage1 from './Stage1';
 import Stage2 from './Stage2';
 import Stage3 from './Stage3';
 import './ChatInterface.css';
 
+const remarkPlugins = [remarkGfm];
+
+// Only memoize user messages (they never change once sent)
+const UserMessage = memo(function UserMessage({ content }) {
+  return (
+    <div className="message-group">
+      <div className="user-message">
+        <div className="message-label">You</div>
+        <div className="message-content">
+          <div className="markdown-content">
+            <ReactMarkdown remarkPlugins={remarkPlugins}>{content}</ReactMarkdown>
+          </div>
+        </div>
+      </div>
+    </div>
+  );
+});
+
+// Memoize completed assistant messages, but skip memo for the active (last) one
+const AssistantMessage = memo(function AssistantMessage({ msg, isActive }) {
+  return (
+    <div className="message-group">
+      <div className="assistant-message">
+        <div className="message-label">LLM Council</div>
+
+        {/* Stage 1 */}
+        {msg.loading?.stage1 && (
+          <div className="stage-loading">
+            <div className="spinner"></div>
+            <span>Running Stage 1: Collecting individual responses...</span>
+          </div>
+        )}
+        {msg.stage1 && <Stage1 responses={msg.stage1} />}
+
+        {/* Stage 2 */}
+        {msg.loading?.stage2 && (
+          <div className="stage-loading">
+            <div className="spinner"></div>
+            <span>Running Stage 2: Peer rankings...</span>
+          </div>
+        )}
+        {msg.stage2 && (
+          <Stage2
+            rankings={msg.stage2}
+            labelToModel={msg.metadata?.label_to_model}
+            aggregateRankings={msg.metadata?.aggregate_rankings}
+          />
+        )}
+
+        {/* Stage 3 */}
+        {msg.loading?.stage3 && (
+          <div className="stage-loading">
+            <div className="spinner"></div>
+            <span>Running Stage 3: Final synthesis...</span>
+          </div>
+        )}
+        {msg.stage3 && <Stage3 finalResponse={msg.stage3} />}
+      </div>
+    </div>
+  );
+}, (prevProps, nextProps) => {
+  // If active (streaming), always re-render
+  if (prevProps.isActive || nextProps.isActive) return false;
+  // Otherwise skip re-render (completed messages don't change)
+  return true;
+});
+
 export default function ChatInterface({
   conversation,
   onSendMessage,
+  onStopGeneration,
   isLoading,
+  pendingInput,
+  onPendingInputConsumed,
 }) {
   const [input, setInput] = useState('');
+  const textareaRef = useRef(null);
   const messagesEndRef = useRef(null);
 
-  const scrollToBottom = () => {
+  useEffect(() => {
     messagesEndRef.current?.scrollIntoView({ behavior: 'smooth' });
-  };
+  }, [conversation, isLoading]);
 
+  // Recover input from stopped generation
   useEffect(() => {
-    scrollToBottom();
-  }, [conversation]);
+    if (pendingInput !== null) {
+      setInput(pendingInput);
+      onPendingInputConsumed();
+      setTimeout(() => textareaRef.current?.focus(), 0);
+    }
+  }, [pendingInput]);
 
   const handleSubmit = (e) => {
     e.preventDefault();
@@ -30,7 +107,6 @@ export default function ChatInterface({
   };
 
   const handleKeyDown = (e) => {
-    // Submit on Enter (without Shift)
     if (e.key === 'Enter' && !e.shiftKey) {
       e.preventDefault();
       handleSubmit(e);
@@ -57,57 +133,13 @@ export default function ChatInterface({
             <p>Ask a question to consult the LLM Council</p>
           </div>
         ) : (
-          conversation.messages.map((msg, index) => (
-            <div key={index} className="message-group">
-              {msg.role === 'user' ? (
-                <div className="user-message">
-                  <div className="message-label">You</div>
-                  <div className="message-content">
-                    <div className="markdown-content">
-                      <ReactMarkdown>{msg.content}</ReactMarkdown>
-                    </div>
-                  </div>
-                </div>
-              ) : (
-                <div className="assistant-message">
-                  <div className="message-label">LLM Council</div>
-
-                  {/* Stage 1 */}
-                  {msg.loading?.stage1 && (
-                    <div className="stage-loading">
-                      <div className="spinner"></div>
-                      <span>Running Stage 1: Collecting individual responses...</span>
-                    </div>
-                  )}
-                  {msg.stage1 && <Stage1 responses={msg.stage1} />}
-
-                  {/* Stage 2 */}
-                  {msg.loading?.stage2 && (
-                    <div className="stage-loading">
-                      <div className="spinner"></div>
-                      <span>Running Stage 2: Peer rankings...</span>
-                    </div>
-                  )}
-                  {msg.stage2 && (
-                    <Stage2
-                      rankings={msg.stage2}
-                      labelToModel={msg.metadata?.label_to_model}
-                      aggregateRankings={msg.metadata?.aggregate_rankings}
-                    />
-                  )}
-
-                  {/* Stage 3 */}
-                  {msg.loading?.stage3 && (
-                    <div className="stage-loading">
-                      <div className="spinner"></div>
-                      <span>Running Stage 3: Final synthesis...</span>
-                    </div>
-                  )}
-                  {msg.stage3 && <Stage3 finalResponse={msg.stage3} />}
-                </div>
-              )}
-            </div>
-          ))
+          conversation.messages.map((msg, index) => {
+            if (msg.role === 'user') {
+              return <UserMessage key={index} content={msg.content} />;
+            }
+            const isLastAssistant = isLoading && index === conversation.messages.length - 1;
+            return <AssistantMessage key={index} msg={msg} isActive={isLastAssistant} />;
+          })
         )}
 
         {isLoading && (
@@ -120,9 +152,9 @@ export default function ChatInterface({
         <div ref={messagesEndRef} />
       </div>
 
-      {conversation.messages.length === 0 && (
-        <form className="input-form" onSubmit={handleSubmit}>
+      <form className="input-form" onSubmit={handleSubmit}>
           <textarea
+            ref={textareaRef}
             className="message-input"
             placeholder="Ask your question... (Shift+Enter for new line, Enter to send)"
             value={input}
@@ -131,15 +163,24 @@ export default function ChatInterface({
             disabled={isLoading}
             rows={3}
           />
-          <button
-            type="submit"
-            className="send-button"
-            disabled={!input.trim() || isLoading}
-          >
-            Send
-          </button>
+          {isLoading ? (
+            <button
+              type="button"
+              className="stop-button"
+              onClick={onStopGeneration}
+            >
+              Stop
+            </button>
+          ) : (
+            <button
+              type="submit"
+              className="send-button"
+              disabled={!input.trim()}
+            >
+              Send
+            </button>
+          )}
         </form>
-      )}
     </div>
   );
 }
diff --git a/frontend/src/components/Stage1.jsx b/frontend/src/components/Stage1.jsx
index 071937c..7478876 100644
--- a/frontend/src/components/Stage1.jsx
+++ b/frontend/src/components/Stage1.jsx
@@ -1,5 +1,6 @@
 import { useState } from 'react';
 import ReactMarkdown from 'react-markdown';
+import remarkGfm from 'remark-gfm';
 import './Stage1.css';
 
 export default function Stage1({ responses }) {
@@ -28,7 +29,7 @@ export default function Stage1({ responses }) {
       <div className="tab-content">
         <div className="model-name">{responses[activeTab].model}</div>
         <div className="response-text markdown-content">
-          <ReactMarkdown>{responses[activeTab].response}</ReactMarkdown>
+          <ReactMarkdown remarkPlugins={[remarkGfm]}>{responses[activeTab].response}</ReactMarkdown>
         </div>
       </div>
     </div>
diff --git a/frontend/src/components/Stage2.jsx b/frontend/src/components/Stage2.jsx
index 2550fa6..5d28ed7 100644
--- a/frontend/src/components/Stage2.jsx
+++ b/frontend/src/components/Stage2.jsx
@@ -1,5 +1,6 @@
 import { useState } from 'react';
 import ReactMarkdown from 'react-markdown';
+import remarkGfm from 'remark-gfm';
 import './Stage2.css';
 
 function deAnonymizeText(text, labelToModel) {
@@ -48,7 +49,7 @@ export default function Stage2({ rankings, labelToModel, aggregateRankings }) {
           {rankings[activeTab].model}
         </div>
         <div className="ranking-content markdown-content">
-          <ReactMarkdown>
+          <ReactMarkdown remarkPlugins={[remarkGfm]}>
             {deAnonymizeText(rankings[activeTab].ranking, labelToModel)}
           </ReactMarkdown>
         </div>
diff --git a/frontend/src/components/Stage3.jsx b/frontend/src/components/Stage3.jsx
index 9a9dbf7..f63a442 100644
--- a/frontend/src/components/Stage3.jsx
+++ b/frontend/src/components/Stage3.jsx
@@ -1,4 +1,5 @@
 import ReactMarkdown from 'react-markdown';
+import remarkGfm from 'remark-gfm';
 import './Stage3.css';
 
 export default function Stage3({ finalResponse }) {
@@ -14,7 +15,7 @@ export default function Stage3({ finalResponse }) {
           Chairman: {finalResponse.model.split('/')[1] || finalResponse.model}
         </div>
         <div className="final-text markdown-content">
-          <ReactMarkdown>{finalResponse.response}</ReactMarkdown>
+          <ReactMarkdown remarkPlugins={[remarkGfm]}>{finalResponse.response}</ReactMarkdown>
         </div>
       </div>
     </div>
diff --git a/frontend/src/index.css b/frontend/src/index.css
index 698f393..b7b4ed7 100644
--- a/frontend/src/index.css
+++ b/frontend/src/index.css
@@ -96,3 +96,25 @@ body {
   border-left: 4px solid #ddd;
   color: #666;
 }
+
+.markdown-content table {
+  border-collapse: collapse;
+  margin: 0 0 12px 0;
+  width: 100%;
+}
+
+.markdown-content th,
+.markdown-content td {
+  border: 1px solid #d0d0d0;
+  padding: 8px 12px;
+  text-align: left;
+}
+
+.markdown-content th {
+  background: #f5f5f5;
+  font-weight: 600;
+}
+
+.markdown-content tr:nth-child(even) {
+  background: #fafafa;
+}