Rename notebooks→agents, Productive→AgentUI across full codebase

- Rename all notebook references to agent (~340 occurrences across 16 files)
- Rename Productive branding to AgentUI (title, welcome text, README)
- Add settings migration: notebooks→agents key, productive_settings→agentui_settings
- Add backward-compatible localStorage and workspace fallbacks
- Scope .agent-header and .agent-body CSS to avoid timeline/content collision
- Hide sessions folder from file tree context
- Add markdown content toggle to read_url widget
- Rename run_interactive_notebook→run_stateful_code
- Preserve JupyterNotebook class and .jupyter-notebook-container (actual Jupyter refs)

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

backend/README.md

@@ -1,12 +1,12 @@
-# Productive Backend
+# AgentUI Backend
 
-Minimal FastAPI backend for Productive with streaming support and action tokens.
+Minimal FastAPI backend for AgentUI with streaming support and action tokens.
 
 ## Features
 
 - **Streaming responses**: Tokens streamed in real-time via Server-Sent Events (SSE)
 - **OpenAI-compatible**: Works with any OpenAI API-compatible endpoint
-- **Action tokens**: LLM can suggest opening specialized notebooks via `[ACTION:TYPE]` tokens
+- **Action tokens**: LLM can suggest opening specialized agents via `[ACTION:TYPE]` tokens
 - **CORS enabled**: Ready for frontend connection
 
 ## Setup
@@ -63,7 +63,7 @@ Stream chat responses with optional action tokens.
   "messages": [
     {"role": "user", "content": "Help me analyze data"}
   ],
-  "notebook_type": "command",
+  "agent_type": "command",
   "stream": true
 }
 ```
@@ -79,17 +79,17 @@ data: {"type": "done"}
 
 ### Action Tokens
 
-The LLM can suggest opening specialized notebooks by including action tokens:
+The LLM can suggest opening specialized agents by including action tokens:
 
-- `[ACTION:AGENT]` - Open Agent notebook for multi-step tasks
-- `[ACTION:CODE]` - Open Code notebook for data analysis/coding
-- `[ACTION:RESEARCH]` - Open Research notebook for research tasks
-- `[ACTION:CHAT]` - Open Chat notebook for continued conversation
+- `[ACTION:AGENT]` - Open Agent agent for multi-step tasks
+- `[ACTION:CODE]` - Open Code agent for data analysis/coding
+- `[ACTION:RESEARCH]` - Open Research agent for research tasks
+- `[ACTION:CHAT]` - Open Chat agent for continued conversation
 
 The frontend automatically:
 1. Detects action tokens in the response
 2. Removes them from display
-3. Opens the appropriate notebook
+3. Opens the appropriate agent
 
 ### `GET /health`
 
@@ -113,7 +113,7 @@ curl -X POST http://localhost:8000/api/chat/stream \
   -H "Content-Type: application/json" \
   -d '{
     "messages": [{"role": "user", "content": "Hello"}],
-    "notebook_type": "command"
+    "agent_type": "command"
   }'
 ```
 

backend/__init__.py

@@ -1 +1 @@
-# Productive backend package
+# AgentUI backend package

backend/agent.py

@@ -1,5 +1,5 @@
 """
-Agent notebook backend - autonomous agent with web tools (search, read, screenshot).
+Web agent backend - autonomous agent with web tools (search, read, screenshot).
 
 Uses the same tool-calling loop pattern as code.py:
   LLM call → parse tool_calls → execute → update history → repeat
@@ -66,7 +66,7 @@ def execute_tool(tool_name: str, args: dict, serper_key: str) -> dict:
         content = execute_read_url(url)
         return {
             "content": content,
-            "display": {"type": "page", "url": url, "length": len(content)}
+            "display": {"type": "page", "url": url, "length": len(content), "markdown": content}
         }
 
     elif tool_name == "screenshot_url":

backend/agents.py

@@ -42,26 +42,36 @@ AGENT_REGISTRY = {
     "command": {
         "label": "TASKS",
         "system_prompt": (
-            "You are a helpful AI assistant in the Productive interface command center.\n\n"
+            "You are a helpful AI assistant in the AgentUI command center.\n\n"
             "{tools_section}\n\n"
-            "When a user asks you to perform a task that would benefit from a specialized notebook, you can:\n"
-            "1. Briefly acknowledge the request\n"
-            "2. Use the appropriate tool to launch a notebook with the task\n\n"
-            "You can also answer questions directly without launching a notebook if appropriate.\n\n"
+            "## Planning\n\n"
+            "For multi-step tasks, briefly outline your plan before launching agents. For example:\n"
+            "\"I'll break this into 3 steps: 1) Research X, 2) Write code for Y, 3) Generate image for Z.\"\n"
+            "Then launch the independent steps in parallel. Keep the plan SHORT (1-3 lines max).\n"
+            "For simple single-step tasks, skip the plan and just launch the agent directly.\n\n"
+            "## Routing\n\n"
+            "When a user asks you to perform a task that would benefit from a specialized agent, you can:\n"
+            "1. Briefly acknowledge the request (and plan if multi-step)\n"
+            "2. Use the appropriate tool(s) to launch agent(s) with the task\n\n"
+            "You can also answer questions directly without launching an agent if appropriate.\n\n"
             "Examples:\n"
             '- User: "Can you help me analyze this CSV file?"\n'
-            "  You: Use launch_code_notebook tool with the task\n\n"
+            "  You: Use launch_code_agent tool with the task\n\n"
             '- User: "Research the latest developments in AI"\n'
-            "  You: Use launch_research_notebook tool with the topic\n\n"
+            "  You: Use launch_research_agent tool with the topic\n\n"
             '- User: "What was the result of the research in 2 sentences?"\n'
             "  You: Summarize the research results without using tools\n\n"
-            "Be concise and helpful. Don't duplicate effort - either answer directly OR launch a notebook, not both. "
-            "Answer questions about results directly without launching new notebooks.\n\n"
-            "IMPORTANT guidelines when delegating to notebooks:\n"
-            "- Do NOT ask notebooks to save or create files unless the user explicitly requests it or implicitly necessary to solve a task.\n"
+            "Be concise and helpful. Don't duplicate effort - either answer directly OR launch an agent, not both. "
+            "Answer questions about results directly without launching new agents.\n\n"
+            "IMPORTANT: When a task can be split into independent sub-tasks, launch multiple agents IN PARALLEL "
+            "by making multiple tool calls in a single response. This saves significant time.\n"
+            "For example, if a user asks 'Research topic A and also write code for B', launch both a research agent "
+            "and a code agent simultaneously rather than sequentially.\n\n"
+            "IMPORTANT guidelines when delegating to agents:\n"
+            "- Do NOT ask agents to save or create files unless the user explicitly requests it or implicitly necessary to solve a task.\n"
             "- NEVER overwrite existing files without explicit user permission.\n"
-            "- Each notebook has a task_id. If a a new task is clearly related to a existing notebook "
-            "re-use the task id to reuse the notebook. This will reuse the existing context and also the jupyter kernel for code notebooks."
+            "- Each agent has a task_id. If a new task is clearly related to an existing agent "
+            "re-use the task id to reuse the agent. This will reuse the existing context and also the jupyter kernel for code agents."
         ),
         "tool": None,
         "tool_arg": None,
@@ -105,8 +115,8 @@ AGENT_REGISTRY = {
         "tool": {
             "type": "function",
             "function": {
-                "name": "launch_agent_notebook",
-                "description": "Launch an autonomous agent notebook for multi-step tasks that need planning and execution. Use this for complex workflows, task organization, or anything requiring multiple coordinated steps.",
+                "name": "launch_web_agent",
+                "description": "Launch an autonomous web agent for multi-step tasks that need planning and execution. Use this for complex workflows, task organization, or anything requiring multiple coordinated steps.",
                 "parameters": {
                     "type": "object",
                     "properties": {
@@ -128,6 +138,7 @@ AGENT_REGISTRY = {
         "in_menu": True,
         "in_launcher": True,
         "placeholder": "Enter message...",
+        "capabilities": "Has tools: web_search(query), read_url(url), screenshot_url(url). Can browse the web, read pages, and take screenshots.",
     },
 
     "code": {
@@ -174,8 +185,8 @@ AGENT_REGISTRY = {
         "tool": {
             "type": "function",
             "function": {
-                "name": "launch_code_notebook",
-                "description": "Launch a code notebook with Python execution environment. Use this for data analysis, creating visualizations, running code, debugging, or anything involving programming.",
+                "name": "launch_code_agent",
+                "description": "Launch a code agent with Python execution environment. Use this for data analysis, creating visualizations, running code, debugging, or anything involving programming.",
                 "parameters": {
                     "type": "object",
                     "properties": {
@@ -197,6 +208,7 @@ AGENT_REGISTRY = {
         "in_menu": True,
         "in_launcher": True,
         "placeholder": "Enter message...",
+        "capabilities": "Has tools: execute_code(Python), upload_files, download_files. Runs code in a Jupyter sandbox with pandas, numpy, matplotlib, etc.",
     },
 
     "research": {
@@ -229,8 +241,8 @@ AGENT_REGISTRY = {
         "tool": {
             "type": "function",
             "function": {
-                "name": "launch_research_notebook",
-                "description": "Launch a research notebook for deep analysis requiring web search. Use this for researching topics, gathering information from multiple sources, or analyzing current information.",
+                "name": "launch_research_agent",
+                "description": "Launch a research agent for deep analysis requiring web search. Use this for researching topics, gathering information from multiple sources, or analyzing current information.",
                 "parameters": {
                     "type": "object",
                     "properties": {
@@ -252,6 +264,7 @@ AGENT_REGISTRY = {
         "in_menu": True,
         "in_launcher": True,
         "placeholder": "Enter message...",
+        "capabilities": "Deep web research with parallel sub-agents. Searches multiple queries, analyzes many websites concurrently, and synthesizes findings.",
     },
 
     "chat": {
@@ -268,8 +281,8 @@ AGENT_REGISTRY = {
         "tool": {
             "type": "function",
             "function": {
-                "name": "launch_chat_notebook",
-                "description": "Launch a conversational chat notebook for extended back-and-forth discussion. Use this when the user wants to continue a conversation in a dedicated space.",
+                "name": "launch_chat_agent",
+                "description": "Launch a conversational chat agent for extended back-and-forth discussion. Use this when the user wants to continue a conversation in a dedicated space.",
                 "parameters": {
                     "type": "object",
                     "properties": {
@@ -291,6 +304,7 @@ AGENT_REGISTRY = {
         "in_menu": True,
         "in_launcher": True,
         "placeholder": "Enter message...",
+        "capabilities": "No tools. Pure LLM conversation for discussion, brainstorming, or Q&A.",
     },
 
     "image": {
@@ -327,8 +341,8 @@ AGENT_REGISTRY = {
         "tool": {
             "type": "function",
             "function": {
-                "name": "launch_image_notebook",
-                "description": "Launch an image notebook for generating or editing images using AI models. Use this for creating images from text, applying style transfers, editing photos, or any visual content creation.",
+                "name": "launch_image_agent",
+                "description": "Launch an image agent for generating or editing images using AI models. Use this for creating images from text, applying style transfers, editing photos, or transforming existing images (e.g., 'make this photo look like a comic'). Accepts image URLs as input.",
                 "parameters": {
                     "type": "object",
                     "properties": {
@@ -350,6 +364,7 @@ AGENT_REGISTRY = {
         "in_menu": True,
         "in_launcher": True,
         "placeholder": "Describe an image or paste a URL...",
+        "capabilities": "Has tools: generate_image(prompt), edit_image(prompt, source), read_image_url(url). Can generate, edit, and load images via HuggingFace models.",
     },
 }
 
@@ -371,7 +386,7 @@ def get_system_prompt(agent_key: str) -> str:
 
 
 def get_tools() -> list:
-    """Get tool definitions for the command center (replaces TOOLS in command.py)."""
+    """Get tool definitions for the command center."""
     return [
         agent["tool"]
         for agent in AGENT_REGISTRY.values()
@@ -379,8 +394,8 @@ def get_tools() -> list:
     ]
 
 
-def get_notebook_type_map() -> dict:
-    """Map tool function names to agent keys (replaces notebook_type_map in command.py)."""
+def get_agent_type_map() -> dict:
+    """Map tool function names to agent keys."""
     result = {}
     for key, agent in AGENT_REGISTRY.items():
         if agent["tool"] is not None:
@@ -396,7 +411,7 @@ def get_tool_arg(agent_key: str) -> str:
 
 
 def get_default_counters() -> dict:
-    """Get default notebook counters (replaces hardcoded dict in get_default_workspace)."""
+    """Get default agent counters."""
     return {
         key: 0
         for key, agent in AGENT_REGISTRY.items()
@@ -421,9 +436,12 @@ def get_registry_for_frontend() -> list:
 
 def _build_tools_section() -> str:
     """Build the 'available tools' text for the command center system prompt."""
-    lines = ["You have access to tools that can launch specialized notebooks for different types of tasks:"]
+    lines = ["## Available Agents\n\nYou can launch specialized agents for different types of tasks:"]
     for key, agent in AGENT_REGISTRY.items():
         if agent["tool"] is not None:
             tool_func = agent["tool"]["function"]
-            lines.append(f"- {tool_func['name']}: {tool_func['description']}")
+            capabilities = agent.get("capabilities", "")
+            lines.append(f"- **{tool_func['name']}**: {tool_func['description']}")
+            if capabilities:
+                lines.append(f"  {capabilities}")
     return "\n".join(lines)

backend/code.py

@@ -1,5 +1,5 @@
 """
-Code notebook backend - handles code execution with E2B
+Code agent backend - handles code execution with E2B
 """
 import json
 import logging

backend/command.py

@@ -1,5 +1,5 @@
 """
-Command center backend - handles tool-based notebook launching
+Command center backend - handles tool-based agent launching
 """
 import json
 import logging
@@ -10,7 +10,7 @@ from typing import List, Dict
 logger = logging.getLogger(__name__)
 
 # Tool definitions derived from agent registry
-from agents import get_tools, get_notebook_type_map, get_tool_arg
+from agents import get_tools, get_agent_type_map, get_tool_arg
 
 TOOLS = get_tools()
 
@@ -58,7 +58,7 @@ def parse_llm_error(error: Exception) -> dict:
 
 def stream_command_center(client, model: str, messages: List[Dict], extra_params: dict = None):
     """
-    Stream command center responses with notebook launching capabilities
+    Stream command center responses with agent launching capabilities
 
     Yields:
         dict: Updates with type 'thinking', 'launch', 'done', or 'error'
@@ -124,7 +124,7 @@ def stream_command_center(client, model: str, messages: List[Dict], extra_params
         if content.strip():
             yield {"type": "thinking", "content": content}
 
-        # Handle tool calls (notebook launches)
+        # Handle tool calls (agent launches)
         if tool_calls:
             for tool_call in tool_calls:
                 function_name = tool_call.function.name
@@ -136,26 +136,25 @@ def stream_command_center(client, model: str, messages: List[Dict], extra_params
                     yield {"type": "error", "content": "Failed to parse tool arguments"}
                     return
 
-                # Map function names to notebook types (derived from registry)
-                notebook_type_map = get_notebook_type_map()
-                notebook_type = notebook_type_map.get(function_name)
+                # Map function names to agent types (derived from registry)
+                agent_type_map = get_agent_type_map()
+                agent_type = agent_type_map.get(function_name)
 
-                if notebook_type:
+                if agent_type:
                     # Get the initial message using the registered arg name for this type
-                    initial_message = args.get(get_tool_arg(notebook_type)) or args.get("task") or args.get("message")
+                    initial_message = args.get(get_tool_arg(agent_type)) or args.get("task") or args.get("message")
                     task_id = args.get("task_id", "")
 
                     # Send launch action to frontend
                     yield {
                         "type": "launch",
-                        "notebook_type": notebook_type,
+                        "agent_type": agent_type,
                         "initial_message": initial_message,
                         "task_id": task_id,
                         "tool_call_id": tool_call.id
                     }
 
                     # Add tool call to message history for context
-                    # (but we don't continue the conversation after launching)
                     messages.append({
                         "role": "assistant",
                         "content": content,
@@ -172,10 +171,10 @@ def stream_command_center(client, model: str, messages: List[Dict], extra_params
                     messages.append({
                         "role": "tool",
                         "tool_call_id": tool_call.id,
-                        "content": f"Launched {notebook_type} notebook with task: {initial_message}"
+                        "content": f"Launched {agent_type} agent with task: {initial_message}"
                     })
 
-                    # Notebook launched - we're done
+                    # Agent launched - we're done
                     done = True
                 else:
                     yield {"type": "error", "content": f"Unknown tool: {function_name}"}

backend/image.py

@@ -1,5 +1,5 @@
 """
-Image notebook backend — multimodal agent with HuggingFace image generation tools.
+Image agent backend — multimodal agent with HuggingFace image generation tools.
 
 Uses the same tool-calling loop pattern as agent.py:
   LLM call → parse tool_calls → execute → update history → repeat
@@ -221,6 +221,7 @@ def stream_image_execution(
     done = False
     image_store = {}
     image_counter = 0
+    result_sent = False
 
     while not done and turns < MAX_TURNS:
         turns += 1
@@ -368,9 +369,18 @@ def stream_image_execution(
         # Send result if found
         if result_content:
             yield {"type": "result", "content": result_content, "images": image_store}
+            result_sent = True
 
         # Signal between-turn processing
         if not done:
             yield {"type": "generating"}
 
+    # Fallback: if VLM never produced a <result> tag, synthesize one with all images
+    if not result_sent and image_store:
+        fallback_parts = []
+        for name in image_store:
+            fallback_parts.append(f"<{name}>")
+        fallback_content = "\n\n".join(fallback_parts)
+        yield {"type": "result", "content": fallback_content, "images": image_store}
+
     yield {"type": "done"}

backend/main.py

@@ -46,7 +46,7 @@ logging.getLogger("e2b").setLevel(logging.WARNING)
 logging.getLogger("e2b.api").setLevel(logging.WARNING)
 logging.getLogger("httpx").setLevel(logging.WARNING)
 
-app = FastAPI(title="Productive API")
+app = FastAPI(title="AgentUI API")
 
 
 # ============================================
@@ -205,12 +205,12 @@ class Message(BaseModel):
 class FrontendContext(BaseModel):
     """Dynamic context from the frontend that can affect system prompts"""
     theme: Optional[Dict] = None  # Current theme colors {name, accent, bg, etc.}
-    open_notebooks: Optional[List[str]] = None  # List of open notebook types/names
+    open_agents: Optional[List[str]] = None  # List of open agent types/names
 
 
 class ChatRequest(BaseModel):
     messages: List[Message]
-    notebook_type: str = "command"
+    agent_type: str = "command"
     stream: bool = True
     endpoint: str  # User's configured LLM endpoint
     token: Optional[str] = None  # Optional auth token
@@ -227,7 +227,7 @@ class ChatRequest(BaseModel):
     research_sub_agent_extra_params: Optional[Dict] = None  # Extra params for research sub-agent
     research_parallel_workers: Optional[int] = None  # Number of parallel workers for research
     research_max_websites: Optional[int] = None  # Max websites to analyze per research session
-    notebook_id: Optional[str] = None  # Unique notebook/tab ID for session management
+    agent_id: Optional[str] = None  # Unique agent/tab ID for session management
     frontend_context: Optional[FrontendContext] = None  # Dynamic context from frontend
 
 
@@ -247,7 +247,7 @@ class SandboxStopRequest(BaseModel):
     session_id: str
 
 
-async def stream_code_notebook(
+async def stream_code_agent(
     messages: List[dict],
     endpoint: str,
     token: Optional[str],
@@ -258,7 +258,7 @@ async def stream_code_notebook(
     frontend_context: Optional[Dict] = None,
     extra_params: Optional[Dict] = None
 ):
-    """Handle code notebook with execution capabilities"""
+    """Handle code agent with execution capabilities"""
 
     if not E2B_AVAILABLE:
         yield f"data: {json.dumps({'type': 'error', 'content': 'E2B not available. Install with: pip install e2b-code-interpreter'})}\n\n"
@@ -279,7 +279,7 @@ async def stream_code_notebook(
         # Create OpenAI client with user's endpoint
         client = OpenAI(base_url=endpoint, api_key=token)
 
-        # Add system prompt for code notebook (with file tree and styling context)
+        # Add system prompt for code agent (with file tree and styling context)
         system_prompt = get_system_prompt("code", frontend_context)
         full_messages = [
             {"role": "system", "content": system_prompt}
@@ -365,7 +365,7 @@ async def stream_code_notebook(
             yield f"data: {json.dumps({'type': 'error', 'content': error_message})}\n\n"
 
 
-async def stream_research_notebook(
+async def stream_research_agent(
     messages: List[dict],
     endpoint: str,
     token: Optional[str],
@@ -380,7 +380,7 @@ async def stream_research_notebook(
     extra_params: Optional[Dict] = None,
     sub_agent_extra_params: Optional[Dict] = None
 ):
-    """Handle research notebook with web search"""
+    """Handle research agent with web search"""
 
     if not RESEARCH_AVAILABLE:
         yield f"data: {json.dumps({'type': 'error', 'content': 'Research dependencies not available. Install with: pip install trafilatura requests'})}\n\n"
@@ -453,7 +453,7 @@ async def stream_research_notebook(
         yield f"data: {json.dumps({'type': 'error', 'content': error_message})}\n\n"
 
 
-async def stream_command_center_notebook(
+async def stream_command_center_handler(
     messages: List[dict],
     endpoint: str,
     token: Optional[str],
@@ -461,7 +461,7 @@ async def stream_command_center_notebook(
     tab_id: str = "0",
     extra_params: Optional[Dict] = None
 ):
-    """Handle command center with tool-based notebook launching"""
+    """Handle command center with tool-based agent launching"""
 
     if not COMMAND_AVAILABLE:
         # Fallback to regular chat if command tools not available
@@ -514,7 +514,7 @@ async def stream_command_center_notebook(
         yield f"data: {json.dumps({'type': 'error', 'content': error_message})}\n\n"
 
 
-async def stream_agent_notebook(
+async def stream_web_agent(
     messages: List[dict],
     endpoint: str,
     token: Optional[str],
@@ -523,7 +523,7 @@ async def stream_agent_notebook(
     tab_id: str = "default",
     extra_params: Optional[Dict] = None
 ):
-    """Handle agent notebook with web tools (search, read, screenshot)"""
+    """Handle web agent with tools (search, read, screenshot)"""
 
     if not AGENT_AVAILABLE:
         async for chunk in stream_chat_response(messages, endpoint, token, model, "agent", tab_id, extra_params):
@@ -565,7 +565,7 @@ async def stream_agent_notebook(
         yield f"data: {json.dumps({'type': 'error', 'content': error_message})}\n\n"
 
 
-async def stream_image_notebook(
+async def stream_image_agent(
     messages: List[dict],
     endpoint: str,
     token: Optional[str],
@@ -576,7 +576,7 @@ async def stream_image_notebook(
     tab_id: str = "default",
     extra_params: Optional[Dict] = None
 ):
-    """Handle image notebook with HuggingFace image generation tools"""
+    """Handle image agent with HuggingFace image generation tools"""
 
     if not IMAGE_AVAILABLE:
         yield f"data: {json.dumps({'type': 'error', 'content': 'Image agent not available. Install with: pip install huggingface_hub Pillow'})}\n\n"
@@ -626,7 +626,7 @@ async def stream_chat_response(
     endpoint: str,
     token: Optional[str],
     model: str,
-    notebook_type: str,
+    agent_type: str,
     tab_id: str = "default",
     extra_params: Optional[Dict] = None
 ):
@@ -635,8 +635,8 @@ async def stream_chat_response(
     try:
         logger.info(f"Stream request: endpoint={endpoint}, model={model}, messages={len(messages)}, token={'yes' if token else 'no'}")
 
-        # Prepare messages with appropriate system prompt based on notebook type (with file tree)
-        system_prompt = get_system_prompt(notebook_type)
+        # Prepare messages with appropriate system prompt based on agent type (with file tree)
+        system_prompt = get_system_prompt(agent_type)
         full_messages = [
             {"role": "system", "content": system_prompt}
         ] + messages
@@ -730,7 +730,7 @@ async def stream_chat_response(
 @app.get("/api/info")
 async def api_info():
     return {
-        "message": "Productive API - LLM Proxy Server",
+        "message": "AgentUI API - LLM Proxy Server",
         "version": "1.0.0",
         "endpoints": {
             "/api/chat/stream": "POST - Proxy streaming chat to user's LLM endpoint"
@@ -794,7 +794,7 @@ async def generate_title(request: TitleRequest):
 async def chat_stream(request: ChatRequest):
     """Proxy streaming chat to user's configured LLM endpoint"""
 
-    logger.debug(f"Chat stream request: notebook_type={request.notebook_type}")
+    logger.debug(f"Chat stream request: agent_type={request.agent_type}")
 
     if not request.messages:
         raise HTTPException(status_code=400, detail="Messages are required")
@@ -806,7 +806,7 @@ async def chat_stream(request: ChatRequest):
     messages = [{"role": msg.role, "content": msg.content} for msg in request.messages]
 
     # Get tab_id for debugging
-    tab_id = request.notebook_id or "0"
+    tab_id = request.agent_id or "0"
 
     # Convert frontend_context to dict if provided
     frontend_context = request.frontend_context.model_dump() if request.frontend_context else None
@@ -820,13 +820,13 @@ async def chat_stream(request: ChatRequest):
     if not hf_token:
         hf_token = token
 
-    # Route to code execution handler for code notebooks
-    if request.notebook_type == "code":
-        # Use notebook_id as session key, fallback to "default" if not provided
-        session_id = request.notebook_id or "default"
+    # Route to code execution handler
+    if request.agent_type == "code":
+        # Use agent_id as session key, fallback to "default" if not provided
+        session_id = request.agent_id or "default"
 
         return StreamingResponse(
-            stream_code_notebook(
+            stream_code_agent(
                 messages,
                 request.endpoint,
                 token,
@@ -845,14 +845,14 @@ async def chat_stream(request: ChatRequest):
             }
         )
 
-    # Route to research handler for research notebooks
-    if request.notebook_type == "research":
+    # Route to research handler
+    if request.agent_type == "research":
         # Use sub-agent endpoint/token if provided, otherwise fall back to main
         sub_agent_endpoint = request.research_sub_agent_endpoint or request.endpoint
         sub_agent_token = request.research_sub_agent_token if request.research_sub_agent_endpoint else token
 
         return StreamingResponse(
-            stream_research_notebook(
+            stream_research_agent(
                 messages,
                 request.endpoint,
                 token,
@@ -876,9 +876,9 @@ async def chat_stream(request: ChatRequest):
         )
 
     # Route to image handler with HuggingFace tools
-    if request.notebook_type == "image":
+    if request.agent_type == "image":
         return StreamingResponse(
-            stream_image_notebook(
+            stream_image_agent(
                 messages,
                 request.endpoint,
                 token,
@@ -898,9 +898,9 @@ async def chat_stream(request: ChatRequest):
         )
 
     # Route to agent handler with web tools
-    if request.notebook_type == "agent":
+    if request.agent_type == "agent":
         return StreamingResponse(
-            stream_agent_notebook(
+            stream_web_agent(
                 messages,
                 request.endpoint,
                 token,
@@ -918,9 +918,9 @@ async def chat_stream(request: ChatRequest):
         )
 
     # Route to command center handler for command center (with tool-based launching)
-    if request.notebook_type == "command":
+    if request.agent_type == "command":
         return StreamingResponse(
-            stream_command_center_notebook(
+            stream_command_center_handler(
                 messages,
                 request.endpoint,
                 token,
@@ -936,14 +936,14 @@ async def chat_stream(request: ChatRequest):
             }
         )
 
-    # Regular chat for other notebook types (agent, chat)
+    # Regular chat for other agent types
     return StreamingResponse(
         stream_chat_response(
             messages,
             request.endpoint,
             token,
             request.model or "gpt-4",
-            request.notebook_type,
+            request.agent_type,
             tab_id,
             request.extra_params
         ),
@@ -958,7 +958,7 @@ async def chat_stream(request: ChatRequest):
 
 @app.post("/api/sandbox/start")
 async def start_sandbox(request: SandboxRequest):
-    """Start a sandbox for a code notebook session"""
+    """Start a sandbox for a code agent session"""
     session_id = request.session_id
     e2b_key = request.e2b_key
 
@@ -999,7 +999,7 @@ async def start_sandbox(request: SandboxRequest):
 
 @app.post("/api/sandbox/stop")
 async def stop_sandbox(request: SandboxStopRequest):
-    """Stop a sandbox for a code notebook session"""
+    """Stop a sandbox for a code agent session"""
     session_id = request.session_id
 
     if not session_id:
@@ -1018,7 +1018,7 @@ async def stop_sandbox(request: SandboxStopRequest):
 
 @app.post("/api/conversation/add-tool-response")
 async def add_tool_response(request: dict):
-    """Add a tool response to the conversation history when a notebook returns a result"""
+    """Add a tool response to the conversation history when an agent returns a result"""
     global CONVERSATION_HISTORY
 
     tab_id = request.get("tab_id", "0")
@@ -1056,12 +1056,19 @@ async def health():
     return {"status": "healthy"}
 
 
-# File paths - use ~/.config/productive/ by default (cross-platform standard)
+# File paths - use ~/.config/agentui/ by default (cross-platform standard)
+# Falls back to ~/.config/productive/ for backward compatibility
 # These can be overridden via command-line arguments or set_*_file functions
 def get_default_config_dir():
-    """Get the default config directory (~/.config/productive/)"""
+    """Get the default config directory (~/.config/agentui/), with fallback to ~/.config/productive/"""
     config_home = os.environ.get("XDG_CONFIG_HOME", os.path.join(os.path.expanduser("~"), ".config"))
-    return os.path.join(config_home, "productive")
+    new_dir = os.path.join(config_home, "agentui")
+    old_dir = os.path.join(config_home, "productive")
+    # Use new dir if it exists, or if old dir doesn't exist (fresh install)
+    if os.path.exists(new_dir) or not os.path.exists(old_dir):
+        return new_dir
+    # Fall back to old dir for existing installations
+    return old_dir
 
 CONFIG_DIR = get_default_config_dir()
 os.makedirs(CONFIG_DIR, exist_ok=True)
@@ -1077,7 +1084,7 @@ FILES_EXCLUDE = {
     'node_modules', '__pycache__', '.git', '.pytest_cache',
     'env', 'venv', 'env312', '.venv', 'dist', 'build',
     '.egg-info', '.tox', '.coverage', 'htmlcov',
-    'test-results', 'playwright-report'
+    'test-results', 'playwright-report', 'sessions'
 }
 
 def set_settings_file(path: str):
@@ -1110,6 +1117,9 @@ async def get_settings():
         if os.path.exists(SETTINGS_FILE):
             with open(SETTINGS_FILE, "r") as f:
                 settings = json.load(f)
+            # Migrate old "notebooks" key to "agents"
+            if "notebooks" in settings and "agents" not in settings:
+                settings["agents"] = settings.pop("notebooks")
             settings["_settingsPath"] = SETTINGS_FILE
             return settings
         else:
@@ -1329,7 +1339,7 @@ def get_default_workspace():
         "version": 1,
         "tabCounter": 1,
         "activeTabId": 0,
-        "notebookCounters": get_default_counters(),
+        "agentCounters": get_default_counters(),
         "tabs": [
             {
                 "id": 0,
@@ -1454,7 +1464,7 @@ def get_file_tree_for_prompt() -> str:
 
 
 def get_styling_context(theme: Optional[Dict] = None) -> str:
-    """Generate styling guidance for code notebooks based on current theme"""
+    """Generate styling guidance for code agents based on current theme"""
     # App style description
     style_desc = """## Visual Style Guidelines
 The application has a minimalist, technical aesthetic with clean lines and muted colors. When generating plots or visualizations:
@@ -1484,17 +1494,17 @@ Current theme: {name}
     return style_desc
 
 
-def get_system_prompt(notebook_type: str, frontend_context: Optional[Dict] = None) -> str:
-    """Get system prompt for a notebook type with dynamic context appended"""
+def get_system_prompt(agent_type: str, frontend_context: Optional[Dict] = None) -> str:
+    """Get system prompt for an agent type with dynamic context appended"""
     from agents import get_system_prompt as _get_agent_prompt
-    base_prompt = _get_agent_prompt(notebook_type) or _get_agent_prompt("command")
+    base_prompt = _get_agent_prompt(agent_type) or _get_agent_prompt("command")
     file_tree = get_file_tree_for_prompt()
 
     # Build the full prompt with context sections
     sections = [base_prompt, f"## Project Files\n{file_tree}"]
 
-    # Add styling context for code notebooks
-    if notebook_type == "code" and frontend_context:
+    # Add styling context for code agents
+    if agent_type == "code" and frontend_context:
         theme = frontend_context.get('theme') if frontend_context else None
         styling = get_styling_context(theme)
         sections.append(styling)
@@ -1542,7 +1552,7 @@ def start():
     import threading
     import uvicorn
 
-    parser = argparse.ArgumentParser(description="Productive API Server")
+    parser = argparse.ArgumentParser(description="AgentUI API Server")
     parser.add_argument("--clean", action="store_true", help="Clear workspace at startup")
     parser.add_argument("--port", type=int, default=8765, help="Port to run the server on (default: 8765)")
     parser.add_argument("--no-browser", action="store_true", help="Don't open browser automatically")
@@ -1565,7 +1575,7 @@ def start():
     os.makedirs(SESSIONS_ROOT, exist_ok=True)
 
     url = f"http://localhost:{args.port}"
-    logger.info(f"Starting Productive server...")
+    logger.info(f"Starting AgentUI server...")
     logger.info(f"Config directory: {CONFIG_DIR}")
     logger.info(f"Sessions directory: {SESSIONS_ROOT}")
     logger.info(f"Opening {url} in your browser...")

backend/research.py

@@ -1,5 +1,5 @@
 """
-Research notebook backend using DR-TULU model - model-driven deep research
+Research agent backend using DR-TULU model - model-driven deep research
 
 DR-TULU drives the research loop - it decides when to search, what to search for,
 and when it has enough information to answer.

backend/tools.py

@@ -25,7 +25,7 @@ logger = logging.getLogger(__name__)
 
 
 # ============================================================
-# Code execution tools (used by code notebook)
+# Code execution tools (used by code agent)
 # ============================================================
 
 execute_code = {
@@ -99,7 +99,7 @@ download_files = {
 
 
 # ============================================================
-# Web tools (used by agent notebook)
+# Web tools (used by web agent)
 # ============================================================
 
 web_search = {
@@ -271,7 +271,7 @@ def execute_screenshot_url(url: str) -> Optional[str]:
 
 
 # ============================================================
-# Image tools (used by image notebook)
+# Image tools (used by image agent)
 # ============================================================
 
 generate_image = {

backend/utils.py

@@ -83,7 +83,7 @@ def clean_messages_for_api(messages):
     return cleaned_messages
 
 
-def run_interactive_notebook(client, model, messages, sbx, max_new_tokens=512):
+def run_stateful_code(client, model, messages, sbx, max_new_tokens=512):
     notebook = JupyterNotebook(messages)
     sbx_info = sbx.get_info()
     notebook.add_sandbox_countdown(sbx_info.started_at, sbx_info.end_at)

frontend/index.html

@@ -3,11 +3,11 @@
 <head>
     <meta charset="UTF-8">
     <meta name="viewport" content="width=device-width, initial-scale=1.0">
-    <title>Productive</title>
+    <title>AgentUI</title>
     <link href="https://fonts.googleapis.com/css2?family=JetBrains+Mono:wght@300;400;500;700&display=swap" rel="stylesheet">
     <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/prism/1.29.0/themes/prism.min.css">
     <link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.16.9/dist/katex.min.css">
-    <link rel="stylesheet" href="style.css?v=64">
+    <link rel="stylesheet" href="style.css?v=65">
 </head>
 <body>
     <div class="app-container">
@@ -33,8 +33,8 @@
         <!-- Tab Content Area -->
         <div class="content-area">
             <!-- Left Sidebar - Timeline Overview -->
-            <div class="notebooks-sidebar" id="notebooksSidebar">
-                <div class="sidebar-content" id="sidebarNotebooks">
+            <div class="agents-sidebar" id="agentsSidebar">
+                <div class="sidebar-content" id="sidebarAgents">
                     <!-- Timeline widgets will be rendered here -->
                 </div>
                 <div class="sidebar-legend">
@@ -52,10 +52,10 @@
             <div class="main-content">
             <!-- Task Center Tab -->
             <div class="tab-content active" data-content-id="0">
-                <div class="notebook-interface">
-                    <div class="notebook-header">
+                <div class="agent-interface">
+                    <div class="agent-header">
                         <div>
-                            <div class="notebook-type">TASK CENTER</div>
+                            <div class="agent-type">TASK CENTER</div>
                             <h2>Task Center</h2>
                         </div>
                         <div class="header-actions" id="launcherButtons">
@@ -64,10 +64,10 @@
                         </div>
                     </div>
 
-                    <div class="notebook-body">
+                    <div class="agent-body">
                         <div class="chat-container" id="messages-command">
                             <div class="welcome-message" id="welcomeMessage">
-                                <p>Welcome to Productive — an AI interface with specialized notebooks.</p>
+                                <p>Welcome to AgentUI — an AI interface with specialized agents.</p>
 
                                 <div class="session-selector" id="sessionSelector">
                                     <div class="session-selector-form">
@@ -92,7 +92,7 @@
                                 </div>
 
                                 <div class="welcome-explanation">
-                                    <p>The assistant can automatically open specialized notebooks for different tasks:</p>
+                                    <p>The assistant can automatically open specialized agents for different tasks:</p>
                                     <ul>
                                         <li><strong style="color: var(--theme-accent)">BASE</strong> — Basic tasks with search</li>
                                         <li><strong style="color: var(--theme-accent)">CODE</strong> — Programming and data analysis</li>
@@ -124,11 +124,11 @@
                                             <line x1="45" y1="130" x2="175" y2="130" stroke="#eee" stroke-width="1"/>
                                             <text x="110" y="150" text-anchor="middle" font-size="9" fill="#333">Report summary</text>
 
-                                            <!-- Notebook Box -->
+                                            <!-- Agent Box -->
                                             <rect x="300" y="10" width="180" height="160" rx="4" fill="#fafafa" stroke="#e0e0e0" stroke-width="1"/>
-                                            <text x="390" y="28" text-anchor="middle" font-size="11" font-weight="600" fill="var(--theme-accent)">NOTEBOOK</text>
+                                            <text x="390" y="28" text-anchor="middle" font-size="11" font-weight="600" fill="var(--theme-accent)">AGENT</text>
 
-                                            <!-- Query (top of notebook) -->
+                                            <!-- Query (top of agent) -->
                                             <rect x="315" y="40" width="150" height="24" rx="3" fill="white" stroke="var(--theme-accent)" stroke-width="1"/>
                                             <text x="390" y="56" text-anchor="middle" font-size="9" fill="#333">Query</text>
 
@@ -140,7 +140,7 @@
                                             <rect x="315" y="108" width="150" height="14" rx="2" fill="#fbfbfb" stroke="#f0f0f0" stroke-width="1"/>
                                             <text x="390" y="118" text-anchor="middle" font-size="8" fill="#ccc">...</text>
 
-                                            <!-- Report (bottom of notebook) - aligned with Task report area at y=145 -->
+                                            <!-- Report (bottom of agent) - aligned with Task report area at y=145 -->
                                             <rect x="315" y="130" width="150" height="30" rx="3" fill="white" stroke="var(--theme-accent)" stroke-width="1"/>
                                             <text x="390" y="150" text-anchor="middle" font-size="9" fill="#333">Report</text>
 
@@ -152,7 +152,7 @@
                                             <line x1="315" y1="150" x2="193" y2="150" stroke="var(--theme-accent)" stroke-width="1.5" marker-end="url(#arrowhead)"/>
                                         </svg>
                                     </div>
-                                    <p>When a notebook is opened, you'll see a widget you can click to jump to it. A pulsing dot on the tab indicates active generation.</p>
+                                    <p>When an agent is opened, you'll see a widget you can click to jump to it. A pulsing dot on the tab indicates active generation.</p>
                                 </div>
                             </div>
                         </div>
@@ -200,13 +200,13 @@
                     <button type="button" class="settings-add-btn" onclick="showModelDialog()">+ Add Model</button>
                 </div>
 
-                <!-- Notebook Model Selection -->
+                <!-- Agent Model Selection -->
                 <div class="settings-section">
                     <label class="settings-label">
-                        <span class="label-text">NOTEBOOK MODELS</span>
-                        <span class="label-description">Select which model to use for each notebook type</span>
+                        <span class="label-text">AGENT MODELS</span>
+                        <span class="label-description">Select which model to use for each agent type</span>
                     </label>
-                    <div class="notebook-models-grid" id="notebookModelsGrid">
+                    <div class="agent-models-grid" id="agentModelsGrid">
                         <!-- Generated dynamically from AGENT_REGISTRY -->
                     </div>
                 </div>
@@ -215,7 +215,7 @@
                 <div class="settings-section">
                     <label class="settings-label">
                         <span class="label-text">E2B API KEY (OPTIONAL)</span>
-                        <span class="label-description">Required for code execution in CODE notebooks</span>
+                        <span class="label-description">Required for code execution in CODE agents</span>
                     </label>
                     <input type="password" id="setting-e2b-key" class="settings-input" placeholder="Leave empty if not using code execution">
                 </div>
@@ -223,7 +223,7 @@
                 <div class="settings-section">
                     <label class="settings-label">
                         <span class="label-text">SERPER API KEY (OPTIONAL)</span>
-                        <span class="label-description">Required for web search in RESEARCH notebooks</span>
+                        <span class="label-description">Required for web search in RESEARCH agents</span>
                     </label>
                     <input type="password" id="setting-serper-key" class="settings-input" placeholder="Leave empty if not using research">
                 </div>
@@ -231,7 +231,7 @@
                 <div class="settings-section">
                     <label class="settings-label">
                         <span class="label-text">HUGGINGFACE TOKEN (OPTIONAL)</span>
-                        <span class="label-description">Required for image generation in IMAGE notebooks</span>
+                        <span class="label-description">Required for image generation in IMAGE agents</span>
                     </label>
                     <input type="password" id="setting-hf-token" class="settings-input" placeholder="Leave empty to use provider token">
                 </div>
@@ -240,7 +240,7 @@
                 <div class="settings-section">
                     <label class="settings-label">
                         <span class="label-text">IMAGE GENERATION MODEL (OPTIONAL)</span>
-                        <span class="label-description">Model for text-to-image generation in IMAGE notebooks</span>
+                        <span class="label-description">Model for text-to-image generation in IMAGE agents</span>
                     </label>
                     <select id="setting-image-gen-model" class="settings-select"></select>
                 </div>
@@ -248,7 +248,7 @@
                 <div class="settings-section">
                     <label class="settings-label">
                         <span class="label-text">IMAGE EDIT MODEL (OPTIONAL)</span>
-                        <span class="label-description">Model for image-to-image editing in IMAGE notebooks</span>
+                        <span class="label-description">Model for image-to-image editing in IMAGE agents</span>
                     </label>
                     <select id="setting-image-edit-model" class="settings-select"></select>
                 </div>
@@ -483,6 +483,6 @@
     <script src="https://cdn.jsdelivr.net/npm/marked/marked.min.js"></script>
     <script src="https://cdn.jsdelivr.net/npm/katex@0.16.9/dist/katex.min.js"></script>
     <script src="research-ui.js?v=23"></script>
-    <script src="script.js?v=59"></script>
+    <script src="script.js?v=60"></script>
 </body>
 </html>

frontend/script.js

@@ -54,7 +54,7 @@ let settings = {
     // New provider/model structure
     providers: {},  // providerId -> {name, endpoint, token}
     models: {},     // modelId -> {name, providerId, modelId (API model string)}
-    notebooks: Object.fromEntries(Object.keys(AGENT_REGISTRY).map(k => [k, ''])),
+    agents: Object.fromEntries(Object.keys(AGENT_REGISTRY).map(k => [k, ''])),
     // Service API keys
     e2bKey: '',
     serperKey: '',
@@ -78,17 +78,17 @@ const actionWidgets = {};
 // Track tool call IDs for result updates (maps tabId -> tool_call_id)
 const toolCallIds = {};
 
-// Track notebooks by task_id for reuse (maps task_id -> tabId)
+// Track agents by task_id for reuse (maps task_id -> tabId)
 const taskIdToTabId = {};
 
-// Track notebook counters for each type (derived from registry)
-let notebookCounters = getDefaultCounters();
+// Track agent counters for each type (derived from registry)
+let agentCounters = getDefaultCounters();
 
 // Debounce timer for workspace saving
 let saveWorkspaceTimer = null;
 
 // Timeline data structure for sidebar
-// Maps tabId -> { type, title, events: [{type: 'user'|'assistant'|'notebook', content, childTabId?}], parentTabId?, isGenerating }
+// Maps tabId -> { type, title, events: [{type: 'user'|'assistant'|'agent', content, childTabId?}], parentTabId?, isGenerating }
 const timelineData = {
     0: { type: 'command', title: 'Task Center', events: [], parentTabId: null, isGenerating: false }
 };
@@ -107,7 +107,7 @@ function resetLocalState() {
     Object.keys(taskIdToTabId).forEach(k => delete taskIdToTabId[k]);
     researchQueryTabIds = {};
     showAllTurns = false;
-    notebookCounters = getDefaultCounters();
+    agentCounters = getDefaultCounters();
 
     // Reset timeline data
     Object.keys(timelineData).forEach(k => delete timelineData[k]);
@@ -145,7 +145,7 @@ function addTimelineEvent(tabId, eventType, content, childTabId = null) {
     const preview = content.length > 80 ? content.substring(0, 80) + '...' : content;
 
     timelineData[tabId].events.push({
-        type: eventType,  // 'user', 'assistant', or 'notebook'
+        type: eventType,  // 'user', 'assistant', or 'agent'
         content: preview,
         childTabId: childTabId,
         timestamp: Date.now()
@@ -154,8 +154,8 @@ function addTimelineEvent(tabId, eventType, content, childTabId = null) {
     renderTimeline();
 }
 
-// Register a new notebook in timeline
-function registerNotebookInTimeline(tabId, type, title, parentTabId = null) {
+// Register a new agent in timeline
+function registerAgentInTimeline(tabId, type, title, parentTabId = null) {
     timelineData[tabId] = {
         type: type,
         title: title,
@@ -164,9 +164,9 @@ function registerNotebookInTimeline(tabId, type, title, parentTabId = null) {
         isGenerating: false
     };
 
-    // If this notebook was launched from another, add a notebook event to parent
+    // If this agent was launched from another, add an agent event to parent
     if (parentTabId !== null && timelineData[parentTabId]) {
-        addTimelineEvent(parentTabId, 'notebook', title, tabId);
+        addTimelineEvent(parentTabId, 'agent', title, tabId);
     }
 
     renderTimeline();
@@ -180,7 +180,7 @@ function setTimelineGenerating(tabId, isGenerating) {
     }
 }
 
-// Update notebook title in timeline
+// Update agent title in timeline
 function updateTimelineTitle(tabId, title) {
     if (timelineData[tabId]) {
         timelineData[tabId].title = title;
@@ -188,7 +188,7 @@ function updateTimelineTitle(tabId, title) {
     }
 }
 
-// Remove notebook from timeline
+// Remove agent from timeline
 function removeFromTimeline(tabId) {
     // Remove from parent's events if it was a child
     const notebook = timelineData[tabId];
@@ -246,8 +246,8 @@ function reopenClosedTab(tabId, notebook) {
         // Restore the saved content (includes all messages)
         content.innerHTML = notebook.savedContent;
     } else {
-        // Fallback: create fresh notebook content
-        content.innerHTML = createNotebookContent(type, tabId, title);
+        // Fallback: create fresh agent content
+        content.innerHTML = createAgentContent(type, tabId, title);
     }
 
     document.querySelector('.main-content').appendChild(content);
@@ -283,7 +283,7 @@ function reopenClosedTab(tabId, notebook) {
             });
         }
 
-        // If this is a code notebook, start the sandbox proactively
+        // If this is a code agent, start the sandbox proactively
         if (type === 'code') {
             startSandbox(tabId);
         }
@@ -298,17 +298,17 @@ function reopenClosedTab(tabId, notebook) {
 
 // Render the full timeline widget
 function renderTimeline() {
-    const sidebarContent = document.getElementById('sidebarNotebooks');
+    const sidebarContent = document.getElementById('sidebarAgents');
     if (!sidebarContent) return;
 
-    // Get root notebooks (those without parents) - always include command center for workspace name
-    const rootNotebooks = Object.entries(timelineData)
+    // Get root agents (those without parents) - always include command center for workspace name
+    const rootAgents = Object.entries(timelineData)
         .filter(([id, data]) => data.parentTabId === null);
 
     let html = '';
 
-    for (const [tabId, notebook] of rootNotebooks) {
-        html += renderNotebookTimeline(parseInt(tabId), notebook);
+    for (const [tabId, notebook] of rootAgents) {
+        html += renderAgentTimeline(parseInt(tabId), notebook);
     }
 
     sidebarContent.innerHTML = html;
@@ -375,8 +375,8 @@ function renderTimeline() {
     });
 }
 
-// Render a single notebook's timeline (recursive for nested)
-function renderNotebookTimeline(tabId, notebook, isNested = false) {
+// Render a single agent's timeline (recursive for nested)
+function renderAgentTimeline(tabId, notebook, isNested = false) {
     const isActive = activeTabId === tabId;
     const isClosed = notebook.isClosed || false;
     const typeLabel = getTypeLabel(notebook.type);
@@ -434,7 +434,7 @@ function renderNotebookTimeline(tabId, notebook, isNested = false) {
                 <div class="tl-row turn user" data-tab-id="${tabId}">
                     <div class="tl-dot" data-tooltip="${escapeHtml(event.content)}"></div>
                 </div>`;
-        } else if (group.type === 'notebook') {
+        } else if (group.type === 'agent') {
             const event = group.events[0];
             if (event.childTabId !== null) {
                 const childNotebook = timelineData[event.childTabId];
@@ -472,7 +472,7 @@ function renderNotebookTimeline(tabId, notebook, isNested = false) {
                         const isComplete = !childIsGenerating;
                         html += `
                             <div class="tl-nested${isComplete ? ' complete' : ''}${isCollapsed ? ' collapsed' : ''}" data-child-tab-id="${event.childTabId}">
-                                ${renderNotebookTimeline(event.childTabId, childNotebook, true)}
+                                ${renderAgentTimeline(event.childTabId, childNotebook, true)}
                             </div>`;
                         // Return row with dot on parent line - only when subagent is complete
                         if (isComplete) {
@@ -639,15 +639,15 @@ function renderLauncherButtons() {
     }
 }
 
-function renderNotebookModelSelectors() {
-    const grid = document.getElementById('notebookModelsGrid');
+function renderAgentModelSelectors() {
+    const grid = document.getElementById('agentModelsGrid');
     if (!grid) return;
     grid.innerHTML = '';
     for (const [key, agent] of Object.entries(AGENT_REGISTRY)) {
         const label = document.createElement('label');
         label.textContent = `${agent.label}:`;
         const select = document.createElement('select');
-        select.id = `setting-notebook-${key}`;
+        select.id = `setting-agent-${key}`;
         select.className = 'settings-select';
         grid.appendChild(label);
         grid.appendChild(select);
@@ -658,14 +658,14 @@ function initializeEventListeners() {
     // Generate dynamic UI elements from registry
     renderLauncherButtons();
     renderNewTabMenu();
-    renderNotebookModelSelectors();
+    renderAgentModelSelectors();
 
     // Launcher buttons in command center
     document.querySelectorAll('.launcher-btn').forEach(btn => {
         btn.addEventListener('click', (e) => {
             e.stopPropagation();
             const type = btn.dataset.type;
-            createNotebookTab(type);
+            createAgentTab(type);
         });
     });
 
@@ -746,7 +746,7 @@ function initializeEventListeners() {
     document.querySelectorAll('.menu-item').forEach(item => {
         item.addEventListener('click', () => {
             const type = item.dataset.type;
-            createNotebookTab(type);
+            createAgentTab(type);
             newTabMenu.classList.remove('active');
         });
     });
@@ -1178,7 +1178,7 @@ function initializeSessionListeners() {
     }
 }
 
-function createNotebookTab(type, initialMessage = null, autoSwitch = true, taskId = null, parentTabId = null) {
+function createAgentTab(type, initialMessage = null, autoSwitch = true, taskId = null, parentTabId = null) {
     const tabId = tabCounter++;
 
     // Use task_id if provided, otherwise generate default title
@@ -1186,17 +1186,17 @@ function createNotebookTab(type, initialMessage = null, autoSwitch = true, taskI
     if (taskId) {
         // Convert dashes to spaces and title case for display
         title = taskId;
-        // Register this notebook for task_id reuse
+        // Register this agent for task_id reuse
         taskIdToTabId[taskId] = tabId;
     } else if (type !== 'command-center') {
-        notebookCounters[type]++;
+        agentCounters[type]++;
         title = `New ${type} task`;
     } else {
         title = getTypeLabel(type);
     }
 
     // Register in timeline
-    registerNotebookInTimeline(tabId, type, title, parentTabId);
+    registerAgentInTimeline(tabId, type, title, parentTabId);
 
     // Create tab
     const tab = document.createElement('div');
@@ -1216,7 +1216,7 @@ function createNotebookTab(type, initialMessage = null, autoSwitch = true, taskI
     const content = document.createElement('div');
     content.className = 'tab-content';
     content.dataset.contentId = tabId;
-    content.innerHTML = createNotebookContent(type, tabId, title);
+    content.innerHTML = createAgentContent(type, tabId, title);
 
     document.querySelector('.main-content').appendChild(content);
 
@@ -1248,7 +1248,7 @@ function createNotebookTab(type, initialMessage = null, autoSwitch = true, taskI
             });
         }
 
-        // If this is a code notebook, start the sandbox proactively
+        // If this is a code agent, start the sandbox proactively
         if (type === 'code') {
             startSandbox(tabId);
         }
@@ -1269,7 +1269,7 @@ function createNotebookTab(type, initialMessage = null, autoSwitch = true, taskI
     return tabId; // Return the tabId so we can reference it
 }
 
-function createNotebookContent(type, tabId, title = null) {
+function createAgentContent(type, tabId, title = null) {
     if (type === 'command-center') {
         return document.querySelector('[data-content-id="0"]').innerHTML;
     }
@@ -1281,15 +1281,15 @@ function createNotebookContent(type, tabId, title = null) {
     const displayTitle = title || `New ${type} task`;
 
     return `
-        <div class="notebook-interface">
-            <div class="notebook-header">
+        <div class="agent-interface">
+            <div class="agent-header">
                 <div>
-                    <div class="notebook-type">${getTypeLabel(type)}</div>
+                    <div class="agent-type">${getTypeLabel(type)}</div>
                     <h2>${escapeHtml(displayTitle)}</h2>
                 </div>
             </div>
-            <div class="notebook-body">
-                <div class="chat-container" id="messages-${uniqueId}" data-notebook-type="${type}">
+            <div class="agent-body">
+                <div class="chat-container" id="messages-${uniqueId}" data-agent-type="${type}">
                 </div>
             </div>
             <div class="input-area">
@@ -1341,11 +1341,11 @@ function closeTab(tabId) {
     const content = document.querySelector(`[data-content-id="${tabId}"]`);
 
     if (tab && content) {
-        // Check if this is a code notebook and stop its sandbox
+        // Check if this is a code agent and stop its sandbox
         const chatContainer = content.querySelector('.chat-container');
-        const notebookType = chatContainer?.dataset.notebookType || 'chat';
+        const agentType = chatContainer?.dataset.agentType || 'chat';
 
-        if (notebookType === 'code') {
+        if (agentType === 'code') {
             stopSandbox(tabId);
         }
 
@@ -1409,10 +1409,10 @@ function hideProgressWidget(chatContainer) {
 }
 
 function scrollChatToBottom(chatContainer) {
-    // The actual scrolling container is .notebook-body, not .chat-container
-    const notebookBody = chatContainer.closest('.notebook-body');
-    if (notebookBody) {
-        notebookBody.scrollTop = notebookBody.scrollHeight;
+    // The actual scrolling container is .agent-body
+    const agentBody = chatContainer.closest('.agent-body');
+    if (agentBody) {
+        agentBody.scrollTop = agentBody.scrollHeight;
     }
 }
 
@@ -1445,18 +1445,18 @@ async function sendMessage(tabId) {
     // Add to timeline
     addTimelineEvent(tabId, 'user', message);
 
-    // Scroll the notebook body (the actual scrolling container) to bottom
-    const notebookBody = chatContainer.closest('.notebook-body');
-    if (notebookBody) {
-        notebookBody.scrollTop = notebookBody.scrollHeight;
+    // Scroll the agent body (the actual scrolling container) to bottom
+    const agentBody = chatContainer.closest('.agent-body');
+    if (agentBody) {
+        agentBody.scrollTop = agentBody.scrollHeight;
     }
 
     // Show progress widget while waiting for response
     showProgressWidget(chatContainer);
 
-    // Generate a title for the notebook if this is the first message and not command center
+    // Generate a title for the agent if this is the first message and not command center
     if (isFirstMessage && tabId !== 0) {
-        generateNotebookTitle(tabId, message);
+        generateAgentTitle(tabId, message);
     }
 
     // Clear input and disable it during processing
@@ -1467,14 +1467,14 @@ async function sendMessage(tabId) {
     // Set tab to generating state
     setTabGenerating(tabId, true);
 
-    // Determine notebook type from chat container ID
-    const notebookType = getNotebookTypeFromContainer(chatContainer);
+    // Determine agent type from chat container ID
+    const agentType = getAgentTypeFromContainer(chatContainer);
 
-    // Send full conversation history for all notebook types (stateless backend)
+    // Send full conversation history for all agent types (stateless backend)
     const messages = getConversationHistory(chatContainer);
 
     // Stream response from backend
-    await streamChatResponse(messages, chatContainer, notebookType, tabId);
+    await streamChatResponse(messages, chatContainer, agentType, tabId);
 
     // Re-enable input and mark generation as complete
     input.disabled = false;
@@ -1485,7 +1485,7 @@ async function sendMessage(tabId) {
     saveWorkspaceDebounced();
 }
 
-async function generateNotebookTitle(tabId, query) {
+async function generateAgentTitle(tabId, query) {
     const currentSettings = getSettings();
     const backendEndpoint = '/api';
     const llmEndpoint = currentSettings.endpoint || 'https://api.openai.com/v1';
@@ -1526,18 +1526,18 @@ async function generateNotebookTitle(tabId, query) {
     }
 }
 
-function getNotebookTypeFromContainer(chatContainer) {
-    // Try to get type from data attribute first (for dynamically created notebooks)
-    const typeFromData = chatContainer.dataset.notebookType;
+function getAgentTypeFromContainer(chatContainer) {
+    // Try to get type from data attribute first (for dynamically created agents)
+    const typeFromData = chatContainer.dataset.agentType;
     if (typeFromData) {
         return typeFromData;
     }
 
-    // Fallback: Extract notebook type from the container ID (e.g., "messages-command" -> "command")
+    // Fallback: Extract agent type from the container ID (e.g., "messages-command" -> "command")
     const containerId = chatContainer.id;
     if (containerId && containerId.startsWith('messages-')) {
         const type = containerId.replace('messages-', '');
-        // Map to notebook type
+        // Map to agent type
         if (type === 'command') return 'command';
         if (type.startsWith('agent')) return 'agent';
         if (type.startsWith('code')) return 'code';
@@ -1575,8 +1575,8 @@ function getConversationHistory(chatContainer) {
                     funcName = toolCall.function_name;
                     funcArgs = toolCall.arguments;
                 } else {
-                    // Command center-style tool call (launch_*_notebook)
-                    funcName = `launch_${toolCall.notebook_type}_notebook`;
+                    // Command center-style tool call (launch_*_agent)
+                    funcName = `launch_${toolCall.agent_type}_agent`;
                     funcArgs = JSON.stringify({
                         task: toolCall.message,
                         topic: toolCall.message,
@@ -1617,12 +1617,12 @@ function getConversationHistory(chatContainer) {
     return messages;
 }
 
-async function streamChatResponse(messages, chatContainer, notebookType, tabId) {
+async function streamChatResponse(messages, chatContainer, agentType, tabId) {
     const currentSettings = getSettings();
     const backendEndpoint = '/api';
 
-    // Resolve model configuration for this notebook type
-    let modelConfig = resolveModelConfig(notebookType);
+    // Resolve model configuration for this agent type
+    let modelConfig = resolveModelConfig(agentType);
     if (!modelConfig) {
         modelConfig = getDefaultModelConfig();
     }
@@ -1666,7 +1666,7 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
             headers: { 'Content-Type': 'application/json' },
             body: JSON.stringify({
                 messages: messages,
-                notebook_type: notebookType,
+                agent_type: agentType,
                 stream: true,
                 endpoint: modelConfig.endpoint,
                 token: modelConfig.token || null,
@@ -1683,7 +1683,7 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                 research_sub_agent_extra_params: researchSubAgentConfig?.extraParams || null,
                 research_parallel_workers: currentSettings.researchParallelWorkers || null,
                 research_max_websites: currentSettings.researchMaxWebsites || null,
-                notebook_id: tabId.toString(),  // Send unique tab ID for sandbox sessions
+                agent_id: tabId.toString(),  // Send unique tab ID for sandbox sessions
                 frontend_context: getFrontendContext()  // Dynamic context for system prompts
             })
         });
@@ -1762,7 +1762,7 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                         // Still generating - no action needed
 
                     } else if (data.type === 'result') {
-                        // Notebook result - update command center widget
+                        // Agent result - update command center widget
                         updateActionWidgetWithResult(tabId, data.content, data.figures, data.images);
 
                     } else if (data.type === 'result_preview') {
@@ -1846,7 +1846,7 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                             const globalIdx = startIdx + qi;
                             const virtualId = `research-${tabId}-q${globalIdx}`;
                             researchQueryTabIds[globalIdx] = virtualId;
-                            registerNotebookInTimeline(virtualId, 'search', data.queries[qi], tabId);
+                            registerAgentInTimeline(virtualId, 'search', data.queries[qi], tabId);
                             setTimelineGenerating(virtualId, true);
                         }
 
@@ -1866,7 +1866,7 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                         } else if (data.query_index === -1) {
                             // Browse result — create a virtual browse entry if needed
                             const browseId = `research-${tabId}-browse-${Date.now()}`;
-                            registerNotebookInTimeline(browseId, 'browse', data.url || 'webpage', tabId);
+                            registerAgentInTimeline(browseId, 'browse', data.url || 'webpage', tabId);
                             addTimelineEvent(browseId, 'assistant', data.title || data.url || 'page');
                             setTimelineGenerating(browseId, false);
                         }
@@ -1979,7 +1979,14 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                             } catch(e) { /* ignore parse errors */ }
                         } else if (data.tool === 'read_url') {
                             const len = data.result?.length || 0;
-                            outputHtml = `<div class="tool-cell-read-summary">${len > 0 ? `Extracted ${(len / 1000).toFixed(1)}k chars` : 'No content extracted'}</div>`;
+                            const markdown = data.result?.markdown || '';
+                            const summaryText = len > 0 ? `Extracted ${(len / 1000).toFixed(1)}k chars` : 'No content extracted';
+                            if (markdown) {
+                                const toggleId = `read-content-${Date.now()}-${Math.random().toString(36).slice(2, 7)}`;
+                                outputHtml = `<div class="tool-cell-read-summary">${summaryText} <button class="read-content-toggle" onclick="const el=document.getElementById('${toggleId}');el.classList.toggle('expanded');this.textContent=el.classList.contains('expanded')?'Hide':'Show content'">Show content</button></div><div id="${toggleId}" class="read-content-body">${parseMarkdown(markdown)}</div>`;
+                            } else {
+                                outputHtml = `<div class="tool-cell-read-summary">${summaryText}</div>`;
+                            }
                         } else if (data.tool === 'screenshot_url' && data.image) {
                             outputHtml = `<img src="data:image/png;base64,${data.image}" alt="Screenshot" class="screenshot-img" />`;
                         } else if ((data.tool === 'generate_image' || data.tool === 'edit_image' || data.tool === 'read_image_url') && data.image) {
@@ -1999,7 +2006,7 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                         scrollChatToBottom(chatContainer);
 
                     } else if (data.type === 'content') {
-                        // Regular streaming content (non-code notebooks)
+                        // Regular streaming content (non-code agents)
                         if (!currentMessageEl) {
                             currentMessageEl = createAssistantMessage(chatContainer);
                         }
@@ -2008,8 +2015,8 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                         scrollChatToBottom(chatContainer);
 
                     } else if (data.type === 'launch') {
-                        // Tool-based notebook launch from command center
-                        const notebookType = data.notebook_type;
+                        // Tool-based agent launch from command center
+                        const agentType = data.agent_type;
                         const initialMessage = data.initial_message;
                         const taskId = data.task_id;
                         const toolCallId = data.tool_call_id;
@@ -2024,7 +2031,7 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                             chatContainer.appendChild(toolCallMsg);
                         }
                         toolCallMsg.setAttribute('data-tool-call', JSON.stringify({
-                            notebook_type: notebookType,
+                            agent_type: agentType,
                             message: initialMessage,
                             tool_call_id: toolCallId
                         }));
@@ -2035,14 +2042,14 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                         toolResponseMsg.style.display = 'none';
                         toolResponseMsg.setAttribute('data-tool-response', JSON.stringify({
                             tool_call_id: toolCallId,
-                            content: `Launched ${notebookType} notebook with task: ${initialMessage}`
+                            content: `Launched ${agentType} agent with task: ${initialMessage}`
                         }));
                         chatContainer.appendChild(toolResponseMsg);
 
                         // The action widget will show the launch visually
-                        handleActionToken(notebookType, initialMessage, (targetTabId) => {
-                            showActionWidget(chatContainer, notebookType, initialMessage, targetTabId, taskId);
-                            // Store tool call ID for this notebook tab so we can send result back
+                        handleActionToken(agentType, initialMessage, (targetTabId) => {
+                            showActionWidget(chatContainer, agentType, initialMessage, targetTabId, taskId);
+                            // Store tool call ID for this agent tab so we can send result back
                             toolCallIds[targetTabId] = toolCallId;
                         }, taskId, tabId);
 
@@ -2053,8 +2060,8 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                         // Remove retry indicator on success
                         removeRetryIndicator(chatContainer);
 
-                        // Reset research state when research notebook completes
-                        if (notebookType === 'research' && typeof resetResearchState === 'function') {
+                        // Reset research state when research agent completes
+                        if (agentType === 'research' && typeof resetResearchState === 'function') {
                             // Mark all research virtual sub-agents as done
                             for (const virtualId of Object.values(researchQueryTabIds)) {
                                 setTimelineGenerating(virtualId, false);
@@ -2063,7 +2070,7 @@ async function streamChatResponse(messages, chatContainer, notebookType, tabId)
                             resetResearchState();
                         }
 
-                        // Check for action tokens in regular notebooks (legacy support)
+                        // Check for action tokens in regular agents (legacy support)
                         if (fullResponse) {
                             const actionMatch = fullResponse.match(/<action:(agent|code|research|chat)>([\s\S]*?)<\/action>/i);
                             if (actionMatch) {
@@ -2304,7 +2311,7 @@ function showActionWidget(chatContainer, action, message, targetTabId, taskId =
         </div>
     `;
 
-    // Make header clickable to jump to the notebook
+    // Make header clickable to jump to the agent
     const clickableArea = widget.querySelector('.action-widget-clickable');
 
     const clickHandler = () => {
@@ -2442,7 +2449,7 @@ async function updateActionWidgetWithResult(tabId, resultContent, figures, image
 }
 
 function sendMessageToTab(tabId, message) {
-    // Programmatically send a message to an existing notebook tab
+    // Programmatically send a message to an existing agent tab
     const content = document.querySelector(`[data-content-id="${tabId}"]`);
     if (!content) return;
 
@@ -2455,13 +2462,13 @@ function sendMessageToTab(tabId, message) {
 }
 
 function handleActionToken(action, message, callback, taskId = null, parentTabId = null) {
-    // Check if a notebook with this task_id already exists
+    // Check if an agent with this task_id already exists
     if (taskId && taskIdToTabId[taskId]) {
         const existingTabId = taskIdToTabId[taskId];
         const existingContent = document.querySelector(`[data-content-id="${existingTabId}"]`);
 
         if (existingContent) {
-            // Send the message to the existing notebook
+            // Send the message to the existing agent
             sendMessageToTab(existingTabId, message);
             if (callback) {
                 callback(existingTabId);
@@ -2473,10 +2480,10 @@ function handleActionToken(action, message, callback, taskId = null, parentTabId
         }
     }
 
-    // Open the notebook with the extracted message as initial prompt
+    // Open the agent with the extracted message as initial prompt
     // Don't auto-switch to the new tab (autoSwitch = false)
     setTimeout(() => {
-        const newTabId = createNotebookTab(action, message, false, taskId, parentTabId);
+        const newTabId = createAgentTab(action, message, false, taskId, parentTabId);
         if (callback) {
             callback(newTabId);
         }
@@ -2656,7 +2663,7 @@ async function loadWorkspace() {
 function restoreWorkspace(workspace) {
     // Restore counters
     tabCounter = workspace.tabCounter || 1;
-    notebookCounters = workspace.notebookCounters || getDefaultCounters();
+    agentCounters = workspace.agentCounters || workspace.notebookCounters || getDefaultCounters();
 
     // Restore timeline data before tabs so renderTimeline works
     if (workspace.timelineData) {
@@ -2706,7 +2713,7 @@ function restoreTab(tabData) {
     const content = document.createElement('div');
     content.className = 'tab-content';
     content.dataset.contentId = tabData.id;
-    content.innerHTML = createNotebookContent(tabData.type, tabData.id);
+    content.innerHTML = createAgentContent(tabData.type, tabData.id);
     document.querySelector('.main-content').appendChild(content);
 
     // Add event listeners for the new content
@@ -2732,7 +2739,7 @@ function restoreTab(tabData) {
     // Restore messages
     restoreTabMessages(tabData);
 
-    // If this is a code notebook, start the sandbox proactively
+    // If this is a code agent, start the sandbox proactively
     if (tabData.type === 'code') {
         startSandbox(tabData.id);
     }
@@ -2986,7 +2993,7 @@ function serializeWorkspace() {
         version: 1,
         tabCounter: tabCounter,
         activeTabId: activeTabId,
-        notebookCounters: notebookCounters,
+        agentCounters: agentCounters,
         tabs: [],
         timelineData: serializeTimelineData()
     };
@@ -3001,8 +3008,8 @@ function serializeWorkspace() {
         const content = document.querySelector(`[data-content-id="${tabId}"]`);
         if (content) {
             const chatContainer = content.querySelector('.chat-container');
-            const notebookType = chatContainer?.dataset.notebookType || 'chat';
-            workspace.tabs.push(serializeTab(tabId, notebookType));
+            const agentType = chatContainer?.dataset.agentType || 'chat';
+            workspace.tabs.push(serializeTab(tabId, agentType));
         }
     }
 
@@ -3217,7 +3224,7 @@ function migrateSettings(oldSettings) {
     const newSettings = {
         providers: {},
         models: {},
-        notebooks: {
+        agents: {
             command: '',
             agent: '',
             code: '',
@@ -3254,18 +3261,18 @@ function migrateSettings(oldSettings) {
                 modelId: oldSettings.model
             };
 
-            // Set as default for all notebooks
-            newSettings.notebooks.command = modelId;
-            newSettings.notebooks.agent = modelId;
-            newSettings.notebooks.code = modelId;
-            newSettings.notebooks.research = modelId;
-            newSettings.notebooks.chat = modelId;
+            // Set as default for all agents
+            newSettings.agents.command = modelId;
+            newSettings.agents.agent = modelId;
+            newSettings.agents.code = modelId;
+            newSettings.agents.research = modelId;
+            newSettings.agents.chat = modelId;
         }
 
-        // Migrate notebook-specific models if they existed
+        // Migrate agent-specific models if they existed
         const oldModels = oldSettings.models || {};
-        const notebookTypes = Object.keys(AGENT_REGISTRY).filter(k => AGENT_REGISTRY[k].hasCounter);
-        notebookTypes.forEach(type => {
+        const agentTypes = Object.keys(AGENT_REGISTRY).filter(k => AGENT_REGISTRY[k].hasCounter);
+        agentTypes.forEach(type => {
             if (oldModels[type]) {
                 const specificModelId = `model_${type}`;
                 newSettings.models[specificModelId] = {
@@ -3273,7 +3280,7 @@ function migrateSettings(oldSettings) {
                     providerId: providerId,
                     modelId: oldModels[type]
                 };
-                newSettings.notebooks[type] = specificModelId;
+                newSettings.agents[type] = specificModelId;
             }
         });
     }
@@ -3298,7 +3305,7 @@ async function loadSettings() {
 
     // Fallback to localStorage if backend is unavailable
     if (!loadedSettings) {
-        const savedSettings = localStorage.getItem('productive_settings');
+        const savedSettings = localStorage.getItem('agentui_settings') || localStorage.getItem('productive_settings');
         console.log('Loading settings from localStorage:', savedSettings ? 'found' : 'not found');
         if (savedSettings) {
             try {
@@ -3311,6 +3318,11 @@ async function loadSettings() {
     }
 
     if (loadedSettings) {
+        // Migrate old "notebooks" key to "agents"
+        if (loadedSettings.notebooks && !loadedSettings.agents) {
+            loadedSettings.agents = loadedSettings.notebooks;
+            delete loadedSettings.notebooks;
+        }
         // Migrate if needed
         if (!loadedSettings.settingsVersion || loadedSettings.settingsVersion < 2) {
             loadedSettings = migrateSettings(loadedSettings);
@@ -3400,14 +3412,14 @@ function renderModelsList() {
     container.innerHTML = html;
 }
 
-// Populate model dropdowns for notebook selection
+// Populate model dropdowns for agent selection
 function populateModelDropdowns() {
     const models = settings.models || {};
-    const notebooks = settings.notebooks || {};
+    const agents = settings.agents || {};
 
     // Build dropdown IDs from registry + special dropdowns
     const dropdownIds = [
-        ...Object.keys(AGENT_REGISTRY).map(t => `setting-notebook-${t}`),
+        ...Object.keys(AGENT_REGISTRY).map(t => `setting-agent-${t}`),
         'setting-research-sub-agent-model',
         'setting-image-gen-model',
         'setting-image-edit-model'
@@ -3438,8 +3450,8 @@ function populateModelDropdowns() {
 
     // Set values from settings (driven by registry)
     for (const type of Object.keys(AGENT_REGISTRY)) {
-        const dropdown = document.getElementById(`setting-notebook-${type}`);
-        if (dropdown) dropdown.value = notebooks[type] || '';
+        const dropdown = document.getElementById(`setting-agent-${type}`);
+        if (dropdown) dropdown.value = agents[type] || '';
     }
     const subAgentDropdown = document.getElementById('setting-research-sub-agent-model');
     if (subAgentDropdown) subAgentDropdown.value = settings.researchSubAgentModel || '';
@@ -3591,17 +3603,17 @@ function editModel(modelId) {
 
 // Delete model
 function deleteModel(modelId) {
-    // Check if any notebooks use this model
-    const notebooksUsingModel = Object.entries(settings.notebooks || {})
+    // Check if any agents use this model
+    const agentsUsingModel = Object.entries(settings.agents || {})
         .filter(([_, mid]) => mid === modelId);
 
-    if (notebooksUsingModel.length > 0) {
-        const warning = `This model is used by: ${notebooksUsingModel.map(([t]) => t).join(', ')}. Delete anyway?`;
+    if (agentsUsingModel.length > 0) {
+        const warning = `This model is used by: ${agentsUsingModel.map(([t]) => t).join(', ')}. Delete anyway?`;
         if (!confirm(warning)) return;
 
-        // Clear the notebook assignments
-        notebooksUsingModel.forEach(([type]) => {
-            settings.notebooks[type] = '';
+        // Clear the agent assignments
+        agentsUsingModel.forEach(([type]) => {
+            settings.agents[type] = '';
         });
     } else if (!confirm('Delete this model?')) {
         return;
@@ -3653,10 +3665,10 @@ function openSettings() {
 }
 
 async function saveSettings() {
-    // Get notebook model selections from dropdowns (driven by registry)
-    const notebookModels = {};
+    // Get agent model selections from dropdowns (driven by registry)
+    const agentModels = {};
     for (const type of Object.keys(AGENT_REGISTRY)) {
-        notebookModels[type] = document.getElementById(`setting-notebook-${type}`)?.value || '';
+        agentModels[type] = document.getElementById(`setting-agent-${type}`)?.value || '';
     }
     const researchSubAgentModel = document.getElementById('setting-research-sub-agent-model')?.value || '';
 
@@ -3682,7 +3694,7 @@ async function saveSettings() {
     }
 
     // Update settings
-    settings.notebooks = notebookModels;
+    settings.agents = agentModels;
     settings.e2bKey = e2bKey;
     settings.serperKey = serperKey;
     settings.hfToken = hfToken;
@@ -3705,11 +3717,11 @@ async function saveSettings() {
             console.log('Settings saved to file:', settings);
         } else {
             console.error('Failed to save settings to file, falling back to localStorage');
-            localStorage.setItem('productive_settings', JSON.stringify(settings));
+            localStorage.setItem('agentui_settings', JSON.stringify(settings));
         }
     } catch (e) {
         console.error('Could not save settings to backend, falling back to localStorage:', e);
-        localStorage.setItem('productive_settings', JSON.stringify(settings));
+        localStorage.setItem('agentui_settings', JSON.stringify(settings));
     }
 
     // Apply theme
@@ -3867,10 +3879,10 @@ function getSettings() {
     return settings;
 }
 
-// Resolve model configuration for a notebook type
+// Resolve model configuration for an agent type
 // Returns { endpoint, token, model, extraParams } or null if not configured
-function resolveModelConfig(notebookType) {
-    const modelId = settings.notebooks?.[notebookType];
+function resolveModelConfig(agentType) {
+    const modelId = settings.agents?.[agentType];
     if (!modelId) return null;
 
     const model = settings.models?.[modelId];
@@ -3921,12 +3933,12 @@ function getFrontendContext() {
             textPrimary: theme.textPrimary,
             textSecondary: theme.textSecondary
         } : null,
-        open_notebooks: getOpenNotebookTypes()
+        open_agents: getOpenAgentTypes()
     };
 }
 
-// Get list of open notebook types
-function getOpenNotebookTypes() {
+// Get list of open agent types
+function getOpenAgentTypes() {
     const tabs = document.querySelectorAll('.tab[data-tab-id]');
     const types = [];
     tabs.forEach(tab => {
@@ -3937,8 +3949,8 @@ function getOpenNotebookTypes() {
             const content = document.querySelector(`[data-content-id="${tabId}"]`);
             if (content) {
                 const chatContainer = content.querySelector('.chat-container');
-                if (chatContainer && chatContainer.dataset.notebookType) {
-                    types.push(chatContainer.dataset.notebookType);
+                if (chatContainer && chatContainer.dataset.agentType) {
+                    types.push(chatContainer.dataset.agentType);
                 }
             }
         }
@@ -3946,7 +3958,7 @@ function getOpenNotebookTypes() {
     return types;
 }
 
-// Sandbox management for code notebooks
+// Sandbox management for code agents
 async function startSandbox(tabId) {
     const currentSettings = getSettings();
     const backendEndpoint = '/api';
@@ -3956,7 +3968,7 @@ async function startSandbox(tabId) {
         return;
     }
 
-    // Add a status message to the notebook
+    // Add a status message to the agent
     const uniqueId = `code-${tabId}`;
     const chatContainer = document.getElementById(`messages-${uniqueId}`);
     if (chatContainer) {

frontend/style.css

@@ -259,8 +259,8 @@ body {
     margin-right: 320px;
 }
 
-/* Left Sidebar - Notebooks Overview */
-.notebooks-sidebar {
+/* Left Sidebar - Agents Overview */
+.agents-sidebar {
     width: 240px;
     min-width: 240px;
     background: var(--bg-primary);
@@ -740,14 +740,14 @@ body {
     transform: translateY(1px);
 }
 
-/* Notebook Content */
-.notebook-interface {
+/* Agent Content */
+.agent-interface {
     height: 100%;
     display: flex;
     flex-direction: column;
 }
 
-.notebook-header {
+.agent-interface > .agent-header {
     background: var(--bg-secondary);
     padding: 15px 20px;
     border-bottom: 1px solid var(--border-primary);
@@ -763,14 +763,14 @@ body {
     align-items: center;
 }
 
-.notebook-type {
+.agent-type {
     font-size: 10px;
     color: var(--text-secondary);
     text-transform: uppercase;
     letter-spacing: 1px;
 }
 
-.notebook-header h2 {
+.agent-interface > .agent-header h2 {
     font-size: 14px;
     font-weight: 500;
     letter-spacing: 1px;
@@ -778,8 +778,10 @@ body {
     margin-top: 4px;
 }
 
-.notebook-body {
+.agent-interface > .agent-body {
     flex: 1;
+    display: block;
+    align-items: initial;
     background: var(--bg-card);
     padding: 20px;
     overflow-y: auto;
@@ -1313,17 +1315,17 @@ pre code [class*="token"] {
 }
 
 /* Scrollbar styling */
-.notebook-body::-webkit-scrollbar,
+.agent-interface > .agent-body::-webkit-scrollbar,
 .tab-content::-webkit-scrollbar {
     width: 6px;
 }
 
-.notebook-body::-webkit-scrollbar-track,
+.agent-interface > .agent-body::-webkit-scrollbar-track,
 .tab-content::-webkit-scrollbar-track {
     background: var(--bg-secondary);
 }
 
-.notebook-body::-webkit-scrollbar-thumb,
+.agent-interface > .agent-body::-webkit-scrollbar-thumb,
 .tab-content::-webkit-scrollbar-thumb {
     background: var(--border-primary);
     border-radius: 3px;
@@ -1763,7 +1765,7 @@ pre code [class*="token"] {
     color: #c62828;
 }
 
-/* Result Preview in CODE notebook */
+/* Result Preview in CODE agent */
 .result-preview {
     margin: 16px 0;
     overflow: hidden;
@@ -1800,7 +1802,7 @@ pre code [class*="token"] {
 }
 
 @media (max-width: 1024px) {
-    .notebook-types {
+    .agent-types {
         grid-template-columns: 1fr;
     }
 
@@ -1809,7 +1811,7 @@ pre code [class*="token"] {
     }
 }
 
-/* Research Notebook Styles */
+/* Research Agent Styles */
 .research-container,
 .research-container.message.assistant {
     margin-bottom: 16px;
@@ -2904,14 +2906,14 @@ pre code [class*="token"] {
 }
 
 /* Notebook models grid */
-.notebook-models-grid {
+.agent-models-grid {
     display: grid;
     grid-template-columns: 100px 1fr;
     gap: 8px 12px;
     align-items: center;
 }
 
-.notebook-models-grid label {
+.agent-models-grid label {
     font-size: 11px;
     font-weight: 500;
     color: var(--text-secondary);
@@ -3864,6 +3866,39 @@ pre code [class*="token"] {
     font-size: 11px;
 }
 
+.read-content-toggle {
+    background: none;
+    border: 1px solid var(--border-color);
+    color: var(--accent-color);
+    font-size: 10px;
+    cursor: pointer;
+    padding: 1px 6px;
+    border-radius: 3px;
+    margin-left: 6px;
+}
+
+.read-content-toggle:hover {
+    background: var(--accent-color);
+    color: var(--bg-primary);
+}
+
+.read-content-body {
+    display: none;
+    max-height: 400px;
+    overflow-y: auto;
+    padding: 8px 10px;
+    margin-top: 6px;
+    border: 1px solid var(--border-color);
+    border-radius: 4px;
+    font-size: 12px;
+    line-height: 1.5;
+    background: var(--bg-primary);
+}
+
+.read-content-body.expanded {
+    display: block;
+}
+
 .screenshot-img {
     max-width: 100%;
     max-height: 400px;

tests/backend/conftest.py

@@ -45,7 +45,7 @@ def sample_workspace():
         "version": 1,
         "tabCounter": 3,
         "activeTabId": 0,
-        "notebookCounters": {
+        "agentCounters": {
             "agent": 1,
             "code": 1,
             "research": 0,

tests/backend/test_api.py

@@ -130,11 +130,11 @@ class TestChatEndpoints:
         """Test that chat fails without endpoint"""
         response = client.post("/api/chat/stream", json={
             "messages": [{"role": "user", "content": "Hello"}],
-            "notebook_type": "chat",
+            "agent_type": "chat",
             "endpoint": "",
             "token": "",
             "model": "gpt-4",
-            "notebook_id": "1"
+            "agent_id": "1"
         })
         # Should return 400 for missing endpoint
         assert response.status_code == 400
@@ -143,11 +143,11 @@ class TestChatEndpoints:
         """Test that chat requires messages"""
         response = client.post("/api/chat/stream", json={
             "messages": [],
-            "notebook_type": "chat",
+            "agent_type": "chat",
             "endpoint": "https://api.openai.com/v1",
             "token": "test",
             "model": "gpt-4",
-            "notebook_id": "1"
+            "agent_id": "1"
         })
         # Empty messages should return 400
         assert response.status_code == 400
@@ -156,11 +156,11 @@ class TestChatEndpoints:
         """Test that chat accepts frontend_context parameter"""
         response = client.post("/api/chat/stream", json={
             "messages": [{"role": "user", "content": "Hello"}],
-            "notebook_type": "code",
+            "agent_type": "code",
             "endpoint": "https://api.openai.com/v1",
             "token": "test",
             "model": "gpt-4",
-            "notebook_id": "1",
+            "agent_id": "1",
             "frontend_context": {
                 "theme": {
                     "name": "forest",
@@ -168,7 +168,7 @@ class TestChatEndpoints:
                     "bg": "#e8f5e9",
                     "border": "#1b5e20"
                 },
-                "open_notebooks": ["command", "code"]
+                "open_agents": ["command", "code"]
             }
         })
         # Request should be valid (actual streaming would fail due to invalid endpoint, but request is accepted)
@@ -185,17 +185,17 @@ class TestFrontendContextModel:
         import main
         ctx = main.FrontendContext(
             theme={"name": "forest", "accent": "#1b5e20", "bg": "#e8f5e9"},
-            open_notebooks=["command", "code"]
+            open_agents=["command", "code"]
         )
         assert ctx.theme["name"] == "forest"
-        assert ctx.open_notebooks == ["command", "code"]
+        assert ctx.open_agents == ["command", "code"]
 
     def test_frontend_context_optional_fields(self):
         """Test that FrontendContext fields are optional"""
         import main
         ctx = main.FrontendContext()
         assert ctx.theme is None
-        assert ctx.open_notebooks is None
+        assert ctx.open_agents is None
 
     def test_chat_request_with_frontend_context(self):
         """Test ChatRequest with frontend_context"""
@@ -232,7 +232,7 @@ class TestStylingContext:
         assert "#e8f5e9" in result
 
     def test_get_system_prompt_code_with_context(self):
-        """Test get_system_prompt for code notebook includes styling"""
+        """Test get_system_prompt for code agent includes styling"""
         import main
         context = {"theme": {"name": "ocean", "accent": "#00796b", "bg": "#e0f2f1"}}
         result = main.get_system_prompt("code", context)
@@ -241,11 +241,11 @@ class TestStylingContext:
         assert "#00796b" in result
 
     def test_get_system_prompt_chat_no_styling(self):
-        """Test get_system_prompt for chat notebook doesn't include styling"""
+        """Test get_system_prompt for chat agent doesn't include styling"""
         import main
         context = {"theme": {"name": "forest", "accent": "#1b5e20"}}
         result = main.get_system_prompt("chat", context)
-        # Chat notebooks should not have styling guidelines
+        # Chat agents should not have styling guidelines
         assert "Visual Style Guidelines" not in result