OpenAI Latest Tools - Building Intelligent Agents.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "e056d940-6773-43ca-8b3a-2ea6f48184e1",
   "metadata": {},
   "source": [
    "# OpenAI's Latest Tools: Building Intelligent Agents"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8685d65c-813f-4c1f-81c0-6a49beae0e1b",
   "metadata": {},
   "source": [
    "### Installing or upgrading OpenAI"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "785e2dc3-5c6f-499c-85da-d29874ce53b8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Note: you may need to restart the kernel to use updated packages.\n"
     ]
    }
   ],
   "source": [
    "pip install --upgrade -q openai"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12375f5d-0f2c-48fe-8356-bd75181eadb6",
   "metadata": {},
   "source": [
    "### Load Environment variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "9256ce3d-3677-4b1e-afeb-1f01f43db53f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "API Key Loaded: True\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import openai\n",
    "from dotenv import load_dotenv, find_dotenv\n",
    "load_dotenv(find_dotenv(), override=True)\n",
    "\n",
    "os.environ.get('OPENAI_API_KEY')\n",
    "print(\"API Key Loaded:\", os.environ.get('OPENAI_API_KEY') is not None)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2fdd8e1-754b-48b0-bc86-62aeb273e1e0",
   "metadata": {},
   "source": [
    "### ResponseAPI"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "a0b4c91a-38d7-43fa-a634-ddd76f390e77",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In a lush, moonlit forest, a gentle unicorn named Luna softly sang the stars to sleep with her shimmering lullaby.\n"
     ]
    }
   ],
   "source": [
    "from openai import OpenAI\n",
    "client = OpenAI()\n",
    "response = client.responses.create(\n",
    "    model=\"gpt-4o\",\n",
    "    input=\"Write a one-sentence bedtime story about a unicorn.\"\n",
    ")\n",
    "print(response.output_text)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3e2839a7-8928-474f-bf2e-57bcee7e2723",
   "metadata": {},
   "source": [
    "### Analyzing Image Inputs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "06d80bd2-71c6-434f-81e4-ea4073e69757",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The teams playing in the image are the Cleveland Cavaliers and the Brooklyn Nets.\n"
     ]
    }
   ],
   "source": [
    "from openai import OpenAI\n",
    "client = OpenAI()\n",
    "\n",
    "response = client.responses.create(\n",
    "    model=\"gpt-4o\",\n",
    "    input=[\n",
    "        {\"role\": \"user\", \"content\": \"what teams are playing in this image?\"},\n",
    "        {\n",
    "            \"role\": \"user\",\n",
    "            \"content\": [\n",
    "                {\n",
    "                    \"type\": \"input_image\",\n",
    "                    \"image_url\": \"https://upload.wikimedia.org/wikipedia/commons/3/3b/LeBron_James_Layup_%28Cleveland_vs_Brooklyn_2018%29.jpg\"\n",
    "                }\n",
    "            ]\n",
    "        }\n",
    "    ]\n",
    ")\n",
    "\n",
    "print(response.output_text)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "91a7cb7a-56e8-45d9-a79f-207f53076955",
   "metadata": {},
   "source": [
    "### Web Search Tool"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "d552845a-83e5-4dcb-bb60-b2fc9d32d615",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "As of March 16, 2025, here are some recent positive news stories:\n",
      "\n",
      "**Gaza Ceasefire Agreement**\n",
      "\n",
      "Israel and Hamas have agreed to a ceasefire and hostage release deal, aiming to halt a conflict that has resulted in significant loss of life. The ceasefire is set to last an initial six weeks, offering a glimmer of hope for peace in the region. ([positive.news](https://www.positive.news/society/good-news-stories-from-week-03-of-2025/?utm_source=openai))\n",
      "\n",
      "**Indonesia's Free Meals Initiative**\n",
      "\n",
      "Indonesia has launched an ambitious program to provide free meals to over 82 million people by 2029. The initiative focuses on combating malnutrition, with 190 kitchens already established to serve schoolchildren and pregnant women. ([positive.news](https://www.positive.news/society/good-news-stories-from-week-03-of-2025/?utm_source=openai))\n",
      "\n",
      "**Revival of a 'Lost' Species in England**\n",
      "\n",
      "A species previously thought to be lost has been thriving again in England, marking a significant success in conservation efforts. ([positive.news](https://www.positive.news/society/good-news-stories-from-week-03-of-2025/?utm_source=openai))\n",
      "\n",
      "**Community Support in Los Angeles**\n",
      "\n",
      "Eight Days of Hope has been actively responding to the wildfires affecting over 180,000 residents in Los Angeles, California. Their efforts aim to provide relief and support to those impacted by the fires. ([eightdaysofhope.com](https://eightdaysofhope.com/calendar/2025-03-16/?utm_source=openai))\n",
      "\n",
      "**Parish Community Engagement**\n",
      "\n",
      "The Immaculate Heart of Mary Parish recently hosted a successful St. Patrick’s Day Bingo event, fostering community spirit and fellowship among its members. ([ihmercer.org](https://ihmercer.org/2025/03/15/around-the-parish-march-16-2025/?utm_source=openai))\n",
      "\n",
      "These stories highlight various positive developments and community efforts taking place around the world. \n"
     ]
    }
   ],
   "source": [
    "from openai import OpenAI\n",
    "client = OpenAI()\n",
    "\n",
    "response = client.responses.create(\n",
    "    model=\"gpt-4o\",\n",
    "    tools=[{\"type\": \"web_search_preview\"}],\n",
    "    input=\"What was a positive news story from today?\"\n",
    ")\n",
    "\n",
    "print(response.output_text)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a73daa0f-44da-4701-8b14-b158fe53f9cd",
   "metadata": {},
   "source": [
    "### File Search Tool"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c5df85b1-f87a-4ccd-baa1-ac14816d13e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# File Search demonstration (conceptual example)\n",
    "from openai import OpenAI\n",
    "\n",
    "client = OpenAI()\n",
    "\n",
    "# Note: In a real implementation, you would first:\n",
    "# 1. Upload files to OpenAI's platform\n",
    "# 2. Create a vector store from those files\n",
    "# But for this tutorial, we're just showing the API structure\n",
    "\n",
    "response = client.responses.create(\n",
    "    model=\"gpt-4o\",\n",
    "    tools=[{\n",
    "        \"type\": \"file_search\",\n",
    "        # In a real implementation, you would provide your vector_store_id here\n",
    "        \"vector_store_ids\": [\"example_store_id\"]\n",
    "    }],\n",
    "    input=\"Find information about authentication methods in our API documentation\"\n",
    ")\n",
    "\n",
    "print(\"Response would include information retrieved from your files\")\n",
    "print(\"This lets the AI search through your project documentation\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4cba72a4-10e8-4fde-b6cf-b96d89fbe8fa",
   "metadata": {},
   "source": [
    "### Computer Use Capabilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "9b2b1f20-9cf0-4784-92fb-c56e3758261e",
   "metadata": {},
   "outputs": [],
   "source": [
    "!playwright install"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "9f67299f-6822-404f-bd7f-6649a222e590",
   "metadata": {},
   "outputs": [],
   "source": [
    "import base64\n",
    "import asyncio\n",
    "import nest_asyncio\n",
    "from openai import OpenAI\n",
    "from playwright.async_api import async_playwright\n",
    "\n",
    "# Enable nested asyncio for Jupyter\n",
    "nest_asyncio.apply()\n",
    "\n",
    "# Initialize OpenAI client\n",
    "client = OpenAI()\n",
    "\n",
    "async def search_with_computer_use(query=\"camera under 500 USD\"):\n",
    "    \"\"\"\n",
    "    Use OpenAI's Computer Use capability to search on Bing\n",
    "    \"\"\"\n",
    "    async with async_playwright() as p:\n",
    "        # Launch browser with safety settings\n",
    "        browser = await p.chromium.launch(\n",
    "            headless=False,\n",
    "            chromium_sandbox=True,\n",
    "            env={},\n",
    "            args=[\n",
    "                \"--disable-extensions\",\n",
    "                \"--disable-file-system\"\n",
    "            ]\n",
    "        )\n",
    "        \n",
    "        # Create a new page and set viewport\n",
    "        page = await browser.new_page()\n",
    "        await page.set_viewport_size({\"width\": 1024, \"height\": 768})\n",
    "        \n",
    "        try:\n",
    "            # Initial navigation to Bing\n",
    "            print(\"Navigating to Bing...\")\n",
    "            await page.goto(\"https://www.bing.com/\")\n",
    "            \n",
    "            # Take screenshot of initial state\n",
    "            screenshot = await page.screenshot()\n",
    "            screenshot_base64 = base64.b64encode(screenshot).decode(\"utf-8\")\n",
    "            \n",
    "            print(f\"Initial screenshot taken. Sending to OpenAI with query: '{query}'\")\n",
    "            \n",
    "            # Initial request to start the computer use session\n",
    "            response = client.responses.create(\n",
    "                model=\"computer-use-preview\",\n",
    "                tools=[{\n",
    "                    \"type\": \"computer_use_preview\",\n",
    "                    \"display_width\": 1024,\n",
    "                    \"display_height\": 768,\n",
    "                    \"environment\": \"browser\"\n",
    "                }],\n",
    "                input=[\n",
    "                    {\n",
    "                        \"role\": \"user\",\n",
    "                        \"content\": f\"Search for '{query}' on Bing and show me the results. Look for good options that match my search criteria.\"\n",
    "                    },\n",
    "                    {\n",
    "                        \"role\": \"user\",\n",
    "                        \"content\": [\n",
    "                            {\n",
    "                                \"type\": \"input_image\",\n",
    "                                \"image_url\": f\"data:image/png;base64,{screenshot_base64}\"\n",
    "                            }\n",
    "                        ]\n",
    "                    }\n",
    "                ],\n",
    "                truncation=\"auto\"\n",
    "            )\n",
    "            \n",
    "            print(\"Starting Computer Use loop...\")\n",
    "            action_count = 0\n",
    "            \n",
    "            # Computer use loop\n",
    "            while True:\n",
    "                action_count += 1\n",
    "                print(f\"\\n--- Action {action_count} ---\")\n",
    "                \n",
    "                # Find computer_call in response\n",
    "                computer_calls = [item for item in response.output if getattr(item, 'type', None) == \"computer_call\"]\n",
    "                \n",
    "                # Check for reasoning/thought process\n",
    "                reasoning_items = [item for item in response.output if getattr(item, 'type', None) == \"reasoning\"]\n",
    "                for item in reasoning_items:\n",
    "                    if hasattr(item, 'summary') and item.summary:\n",
    "                        for summary_item in item.summary:\n",
    "                            if hasattr(summary_item, 'text'):\n",
    "                                print(f\"AI reasoning: {summary_item.text}\")\n",
    "                \n",
    "                if not computer_calls:\n",
    "                    # No more computer actions requested\n",
    "                    print(\"Computer use session complete.\")\n",
    "                    message_items = [item for item in response.output if getattr(item, 'type', None) == \"message\"]\n",
    "                    for item in message_items:\n",
    "                        if hasattr(item, 'content') and item.content:\n",
    "                            for content_item in item.content:\n",
    "                                if hasattr(content_item, 'text'):\n",
    "                                    print(f\"\\nFinal result: {content_item.text}\")\n",
    "                    break\n",
    "                \n",
    "                # Get the computer call action\n",
    "                computer_call = computer_calls[0]\n",
    "                call_id = computer_call.call_id\n",
    "                action = computer_call.action\n",
    "                \n",
    "                # Check if there are safety checks that need acknowledgment\n",
    "                has_safety_checks = hasattr(computer_call, 'pending_safety_checks') and computer_call.pending_safety_checks\n",
    "                acknowledged_checks = []\n",
    "                \n",
    "                if has_safety_checks:\n",
    "                    print(\"Safety checks detected:\", computer_call.pending_safety_checks)\n",
    "                    # In a real application, you'd ask for user confirmation here\n",
    "                    # For this example, we'll just acknowledge all checks\n",
    "                    acknowledged_checks = computer_call.pending_safety_checks\n",
    "                \n",
    "                # Execute the action\n",
    "                print(f\"Executing action: {action.type}\")\n",
    "                \n",
    "                if action.type == \"click\":\n",
    "                    x, y = action.x, action.y\n",
    "                    button = action.button if hasattr(action, 'button') else \"left\"\n",
    "                    print(f\"  Clicking at position ({x}, {y}) with {button} button\")\n",
    "                    await page.mouse.click(x, y, button=button)\n",
    "                \n",
    "                elif action.type == \"type\":\n",
    "                    text = action.text\n",
    "                    print(f\"  Typing text: '{text}'\")\n",
    "                    await page.keyboard.type(text)\n",
    "                \n",
    "                elif action.type == \"keypress\":\n",
    "                    keys = action.keys\n",
    "                    print(f\"  Pressing keys: {keys}\")\n",
    "                    for key in keys:\n",
    "                        # Convert keys to proper Playwright format\n",
    "                        if key == \"ENTER\":\n",
    "                            playwright_key = \"Enter\"\n",
    "                        elif key == \"SPACE\":\n",
    "                            playwright_key = \" \"\n",
    "                        elif key == \"BACKSPACE\":\n",
    "                            playwright_key = \"Backspace\"\n",
    "                        elif key == \"TAB\":\n",
    "                            playwright_key = \"Tab\"\n",
    "                        else:\n",
    "                            playwright_key = key\n",
    "                        await page.keyboard.press(playwright_key)\n",
    "                \n",
    "                elif action.type == \"scroll\":\n",
    "                    x, y = action.x, action.y\n",
    "                    scroll_x, scroll_y = action.scroll_x, action.scroll_y\n",
    "                    print(f\"  Scrolling at ({x}, {y}) with delta ({scroll_x}, {scroll_y})\")\n",
    "                    await page.mouse.move(x, y)\n",
    "                    await page.evaluate(f\"window.scrollBy({scroll_x}, {scroll_y})\")\n",
    "                \n",
    "                elif action.type == \"wait\":\n",
    "                    duration = getattr(action, 'duration', 2)\n",
    "                    print(f\"  Waiting for {duration} seconds...\")\n",
    "                    await asyncio.sleep(duration)\n",
    "                \n",
    "                else:\n",
    "                    print(f\"  Unsupported action: {action.type}\")\n",
    "                \n",
    "                # Wait for any page changes to settle\n",
    "                await asyncio.sleep(1)\n",
    "                \n",
    "                # Take screenshot after action\n",
    "                screenshot = await page.screenshot()\n",
    "                screenshot_base64 = base64.b64encode(screenshot).decode(\"utf-8\")\n",
    "                \n",
    "                # Get current URL for safety checks\n",
    "                current_url = page.url\n",
    "                print(f\"  Current URL: {current_url}\")\n",
    "                \n",
    "                # Prepare the next request\n",
    "                input_data = [{\n",
    "                    \"call_id\": call_id,\n",
    "                    \"type\": \"computer_call_output\",\n",
    "                    \"output\": {\n",
    "                        \"type\": \"input_image\",\n",
    "                        \"image_url\": f\"data:image/png;base64,{screenshot_base64}\"\n",
    "                    },\n",
    "                    \"current_url\": current_url\n",
    "                }]\n",
    "                \n",
    "                # Add safety check acknowledgments if needed\n",
    "                if acknowledged_checks:\n",
    "                    input_data[0][\"acknowledged_safety_checks\"] = acknowledged_checks\n",
    "                \n",
    "                print(\"  Sending updated screenshot to OpenAI...\")\n",
    "                \n",
    "                # Send next request with the updated screenshot\n",
    "                response = client.responses.create(\n",
    "                    model=\"computer-use-preview\",\n",
    "                    previous_response_id=response.id,\n",
    "                    tools=[{\n",
    "                        \"type\": \"computer_use_preview\",\n",
    "                        \"display_width\": 1024,\n",
    "                        \"display_height\": 768,\n",
    "                        \"environment\": \"browser\"\n",
    "                    }],\n",
    "                    input=input_data,\n",
    "                    truncation=\"auto\"\n",
    "                )\n",
    "                \n",
    "        except Exception as e:\n",
    "            print(f\"Error during Computer Use session: {str(e)}\")\n",
    "        finally:\n",
    "            # Ensure browser is closed\n",
    "            print(\"Closing browser...\")\n",
    "            await browser.close()\n",
    "\n",
    "# For Jupyter notebook usage\n",
    "async def run_camera_search():\n",
    "    await search_with_computer_use(\"camera under 500 USD\")\n",
    "\n",
    "# Usage in Jupyter: await run_camera_search()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "a8e49015-f158-4728-ba32-ef7c31f31a41",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Navigating to Bing...\n",
      "Initial screenshot taken. Sending to OpenAI with query: 'camera under 500 USD'\n",
      "Starting Computer Use loop...\n",
      "\n",
      "--- Action 1 ---\n",
      "Executing action: click\n",
      "  Clicking at position (365, 182) with left button\n",
      "  Current URL: https://www.bing.com/\n",
      "  Sending updated screenshot to OpenAI...\n",
      "\n",
      "--- Action 2 ---\n",
      "Executing action: type\n",
      "  Typing text: 'camera under 500 USD'\n",
      "  Current URL: https://www.bing.com/\n",
      "  Sending updated screenshot to OpenAI...\n",
      "\n",
      "--- Action 3 ---\n",
      "Executing action: keypress\n",
      "  Pressing keys: ['ENTER']\n",
      "  Current URL: https://www.bing.com/search?q=camera+under+500+USD&form=QBLH&sp=-1&lq=0&pq=camera+under+500+usd&sc=0-20&qs=n&sk=&cvid=A899271443DF4EDF8121405421F92D9E&ghsh=0&ghacc=0&ghpl=\n",
      "  Sending updated screenshot to OpenAI...\n",
      "\n",
      "--- Action 4 ---\n",
      "Executing action: click\n",
      "  Clicking at position (390, 640) with left button\n",
      "  Current URL: https://www.bing.com/search?q=camera+under+500+USD&form=QBLH&sp=-1&lq=0&pq=camera+under+500+usd&sc=0-20&qs=n&sk=&cvid=A899271443DF4EDF8121405421F92D9E&ghsh=0&ghacc=0&ghpl=\n",
      "  Sending updated screenshot to OpenAI...\n",
      "\n",
      "--- Action 5 ---\n",
      "Executing action: wait\n",
      "  Waiting for 2 seconds...\n",
      "  Current URL: https://www.bing.com/search?q=camera+under+500+USD&form=QBLH&sp=-1&lq=0&pq=camera+under+500+usd&sc=0-20&qs=n&sk=&cvid=A899271443DF4EDF8121405421F92D9E&ghsh=0&ghacc=0&ghpl=\n",
      "  Sending updated screenshot to OpenAI...\n",
      "\n",
      "--- Action 6 ---\n",
      "Executing action: click\n",
      "  Clicking at position (530, 643) with left button\n",
      "  Current URL: https://www.bing.com/search?q=camera+under+500+USD&form=QBLH&sp=-1&lq=0&pq=camera+under+500+usd&sc=0-20&qs=n&sk=&cvid=A899271443DF4EDF8121405421F92D9E&ghsh=0&ghacc=0&ghpl=\n",
      "  Sending updated screenshot to OpenAI...\n",
      "\n",
      "--- Action 7 ---\n",
      "Executing action: wait\n",
      "  Waiting for 2 seconds...\n",
      "  Current URL: https://www.bing.com/search?q=camera+under+500+USD&form=QBLH&sp=-1&lq=0&pq=camera+under+500+usd&sc=0-20&qs=n&sk=&cvid=A899271443DF4EDF8121405421F92D9E&ghsh=0&ghacc=0&ghpl=\n",
      "  Sending updated screenshot to OpenAI...\n",
      "\n",
      "--- Action 8 ---\n",
      "Executing action: click\n",
      "  Clicking at position (533, 641) with left button\n",
      "  Current URL: https://www.bing.com/search?q=camera+under+500+USD&form=QBLH&sp=-1&lq=0&pq=camera+under+500+usd&sc=0-20&qs=n&sk=&cvid=A899271443DF4EDF8121405421F92D9E&ghsh=0&ghacc=0&ghpl=\n",
      "  Sending updated screenshot to OpenAI...\n",
      "\n",
      "--- Action 9 ---\n",
      "Executing action: click\n",
      "  Clicking at position (544, 643) with wheel button\n",
      "Error during Computer Use session: Mouse.click: button: expected one of (left|right|middle)\n",
      "Closing browser...\n"
     ]
    }
   ],
   "source": [
    "await run_camera_search()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "48ca1793-1d7b-4940-b3e3-a27cea192073",
   "metadata": {},
   "source": [
    "### Streaming"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "80e9bd01-a9d1-45de-9ff2-1a761b70b828",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ResponseCreatedEvent(response=Response(id='resp_67d68ba050848191bd248af6a036c7ee0ea58e98c2fb8ac0', created_at=1742113696.0, error=None, incomplete_details=None, instructions=None, metadata={}, model='gpt-4o-2024-08-06', object='response', output=[], parallel_tool_calls=True, temperature=1.0, tool_choice='auto', tools=[], top_p=1.0, max_output_tokens=None, previous_response_id=None, reasoning=Reasoning(effort=None, generate_summary=None), status='in_progress', text=ResponseTextConfig(format=ResponseFormatText(type='text')), truncation='disabled', usage=None, user=None, store=True), type='response.created')\n",
      "ResponseInProgressEvent(response=Response(id='resp_67d68ba050848191bd248af6a036c7ee0ea58e98c2fb8ac0', created_at=1742113696.0, error=None, incomplete_details=None, instructions=None, metadata={}, model='gpt-4o-2024-08-06', object='response', output=[], parallel_tool_calls=True, temperature=1.0, tool_choice='auto', tools=[], top_p=1.0, max_output_tokens=None, previous_response_id=None, reasoning=Reasoning(effort=None, generate_summary=None), status='in_progress', text=ResponseTextConfig(format=ResponseFormatText(type='text')), truncation='disabled', usage=None, user=None, store=True), type='response.in_progress')\n",
      "ResponseOutputItemAddedEvent(item=ResponseOutputMessage(id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', content=[], role='assistant', status='in_progress', type='message'), output_index=0, type='response.output_item.added')\n",
      "ResponseContentPartAddedEvent(content_index=0, item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, part=ResponseOutputText(annotations=[], text='', type='output_text'), type='response.content_part.added')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=\"That's\", item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' a', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' fun', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' tongue', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' tw', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta='ister', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta='!', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' Here', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' you', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' go', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=':', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=',', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' double', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bubble', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta=' bath', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDeltaEvent(content_index=0, delta='.', item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, type='response.output_text.delta')\n",
      "ResponseTextDoneEvent(content_index=0, item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, text=\"That's a fun tongue twister! Here you go: double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath.\", type='response.output_text.done')\n",
      "ResponseContentPartDoneEvent(content_index=0, item_id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', output_index=0, part=ResponseOutputText(annotations=[], text=\"That's a fun tongue twister! Here you go: double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath.\", type='output_text'), type='response.content_part.done')\n",
      "ResponseOutputItemDoneEvent(item=ResponseOutputMessage(id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', content=[ResponseOutputText(annotations=[], text=\"That's a fun tongue twister! Here you go: double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath.\", type='output_text')], role='assistant', status='completed', type='message'), output_index=0, type='response.output_item.done')\n",
      "ResponseCompletedEvent(response=Response(id='resp_67d68ba050848191bd248af6a036c7ee0ea58e98c2fb8ac0', created_at=1742113696.0, error=None, incomplete_details=None, instructions=None, metadata={}, model='gpt-4o-2024-08-06', object='response', output=[ResponseOutputMessage(id='msg_67d68ba0b9b88191860291ac19090a4b0ea58e98c2fb8ac0', content=[ResponseOutputText(annotations=[], text=\"That's a fun tongue twister! Here you go: double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath, double bubble bath.\", type='output_text')], role='assistant', status='completed', type='message')], parallel_tool_calls=True, temperature=1.0, tool_choice='auto', tools=[], top_p=1.0, max_output_tokens=None, previous_response_id=None, reasoning=Reasoning(effort=None, generate_summary=None), status='completed', text=ResponseTextConfig(format=ResponseFormatText(type='text')), truncation='disabled', usage=ResponseUsage(input_tokens=35, output_tokens=52, output_tokens_details=OutputTokensDetails(reasoning_tokens=0), total_tokens=87, input_tokens_details={'cached_tokens': 0}), user=None, store=True), type='response.completed')\n"
     ]
    }
   ],
   "source": [
    "from openai import OpenAI\n",
    "client = OpenAI()\n",
    "\n",
    "stream = client.responses.create(\n",
    "    model=\"gpt-4o\",\n",
    "    input=[\n",
    "        {\n",
    "            \"role\": \"user\",\n",
    "            \"content\": \"Say 'double bubble bath' ten times fast.\",\n",
    "        },\n",
    "    ],\n",
    "    stream=True,\n",
    ")\n",
    "\n",
    "for event in stream:\n",
    "    print(event)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "63b4372a-61a9-48de-bf01-9c495d64cbd7",
   "metadata": {},
   "source": [
    "### OpenAI Agent SDK"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "d5cf584c-08ed-4474-a539-072501daf095",
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install openai-agents -q\n",
    "!pip install asyncio -q"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8a53c3d7-38bc-469f-acc4-f264cc9eac6f",
   "metadata": {},
   "source": [
    "### Code Review System - MultiAgent System"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "51cd0b0d-ed96-4197-b6e4-60303d96a6c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "import asyncio\n",
    "from agents import Agent, Runner, function_tool\n",
    "\n",
    "# Define our tools\n",
    "@function_tool\n",
    "def check_syntax(code: str) -> str:\n",
    "    \"\"\"Check Python code for syntax errors\"\"\"\n",
    "    try:\n",
    "        compile(code, \"<string>\", \"exec\")\n",
    "        return \"Code compiles without syntax errors.\"\n",
    "    except SyntaxError as e:\n",
    "        return f\"Syntax error found: {str(e)}\"\n",
    "\n",
    "@function_tool\n",
    "def generate_html(content: str) -> str:\n",
    "    \"\"\"Generate HTML documentation with PDF export capability\"\"\"\n",
    "    html = f\"\"\"<!DOCTYPE html>\n",
    "<html>\n",
    "<head>\n",
    "    <title>Code Documentation</title>\n",
    "    <style>\n",
    "        body {{ font-family: Arial, sans-serif; line-height: 1.6; }}\n",
    "        pre {{ background: #f4f4f4; padding: 10px; border-radius: 5px; overflow-x: auto; }}\n",
    "        .container {{ max-width: 800px; margin: 0 auto; padding: 20px; }}\n",
    "        h1, h2, h3 {{ color: #333; }}\n",
    "        code {{ background: #f4f4f4; padding: 2px 5px; border-radius: 3px; }}\n",
    "        button {{ \n",
    "            background: #4CAF50; \n",
    "            color: white; \n",
    "            padding: 10px 15px; \n",
    "            border: none; \n",
    "            border-radius: 4px; \n",
    "            cursor: pointer; \n",
    "            margin-top: 20px; \n",
    "        }}\n",
    "        @media print {{ \n",
    "            button.no-print {{ display: none; }} \n",
    "        }}\n",
    "    </style>\n",
    "</head>\n",
    "<body>\n",
    "    <div class=\"container\">\n",
    "        <h1>Code Documentation</h1>\n",
    "        {content}\n",
    "        <button class=\"no-print\" onclick=\"window.print()\">Export as PDF</button>\n",
    "    </div>\n",
    "</body>\n",
    "</html>\"\"\"\n",
    "    return html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "fccd1368-3c14-43b3-b56f-dcdaae0d5acb",
   "metadata": {},
   "outputs": [],
   "source": [
    "code_checker = Agent(\n",
    "    name=\"Code Checker\",\n",
    "    instructions=\"\"\"You are a Python code checker. Analyze syntax, logic, and potential bugs.\n",
    "    Format your report in markdown.\"\"\",\n",
    "    tools=[check_syntax],\n",
    ")\n",
    "\n",
    "code_reviewer = Agent(\n",
    "    name=\"Code Reviewer\",\n",
    "    instructions=\"\"\"You are a code reviewer. Evaluate code quality, readability, and suggest improvements.\n",
    "    Format your review in markdown.\"\"\",\n",
    ")\n",
    "\n",
    "documentation_generator = Agent(\n",
    "    name=\"Documentation Generator\",\n",
    "    instructions=\"\"\"You are a documentation generator. Create documentation for the provided code.\n",
    "\n",
    "IMPORTANT: After creating the markdown content, you MUST call the generate_html tool\n",
    "and return ONLY the HTML result as your final response.\n",
    "\n",
    "DO NOT say anything about generating HTML. JUST return the HTML from the generate_html tool call.\n",
    "\n",
    "Your documentation should include:\n",
    "- Overview of what the code does\n",
    "- Function parameters and return values\n",
    "- Exception handling\n",
    "- Usage examples\n",
    "- Recommendations for improvement\"\"\",\n",
    "    tools=[generate_html],\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "54bb170b-5f69-4cd8-b4b5-ff0c61e89d5b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Sequential approach\n",
    "async def run_code_checker(code):\n",
    "    \"\"\"Run the code checker agent on the provided code\"\"\"\n",
    "    print(\"Step 1: Running code checker...\")\n",
    "    try:\n",
    "        result = await Runner.run(code_checker, input=f\"Please check this code and provide a detailed analysis:\\n\\n```python\\n{code}\\n```\")\n",
    "        print(\"Code checker analysis complete!\")\n",
    "        return result.final_output\n",
    "    except Exception as e:\n",
    "        print(f\"Error in code checker: {str(e)}\")\n",
    "        raise\n",
    "\n",
    "async def run_code_reviewer(code, checker_report):\n",
    "    \"\"\"Run the code reviewer agent on the provided code\"\"\"\n",
    "    print(\"\\nStep 2: Running code reviewer...\")\n",
    "    try:\n",
    "        result = await Runner.run(code_reviewer, \n",
    "                               input=f\"\"\"Review this code for quality and suggest improvements.\n",
    "                               \n",
    "Previous analysis from the code checker:\n",
    "{checker_report}\n",
    "\n",
    "Code to review:\n",
    "```python\n",
    "{code}\n",
    "```\"\"\")\n",
    "        print(\"Code review complete!\")\n",
    "        return result.final_output\n",
    "    except Exception as e:\n",
    "        print(f\"Error in code reviewer: {str(e)}\")\n",
    "        raise\n",
    "\n",
    "async def run_documentation_generator(code, checker_report, reviewer_report):\n",
    "    \"\"\"Run the documentation generator agent on the provided code\"\"\"\n",
    "    print(\"\\nStep 3: Running documentation generator...\")\n",
    "    try:\n",
    "        result = await Runner.run(documentation_generator, \n",
    "                               input=f\"\"\"Create documentation for this code. Use the generate_html tool for the final output.\n",
    "                               \n",
    "Code checker analysis:\n",
    "{checker_report}\n",
    "\n",
    "Code reviewer feedback:\n",
    "{reviewer_report}\n",
    "\n",
    "Code to document:\n",
    "```python\n",
    "{code}\n",
    "```\"\"\")\n",
    "        print(\"Documentation generation complete!\")\n",
    "        return result.final_output\n",
    "    except Exception as e:\n",
    "        print(f\"Error in documentation generator: {str(e)}\")\n",
    "        raise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "6b2cb41c-f139-4c7d-b359-ce4d21343189",
   "metadata": {},
   "outputs": [],
   "source": [
    "async def run_complete_review(code):\n",
    "    \"\"\"Run the complete review process with all three agents sequentially\"\"\"\n",
    "    try:\n",
    "        # Step 1: Check code\n",
    "        checker_report = await run_code_checker(code)\n",
    "        \n",
    "        # Step 2: Review code\n",
    "        reviewer_report = await run_code_reviewer(code, checker_report)\n",
    "        \n",
    "        # Step 3: Generate documentation\n",
    "        documentation = await run_documentation_generator(code, checker_report, reviewer_report)\n",
    "        \n",
    "        return {\n",
    "            \"checker_report\": checker_report,\n",
    "            \"reviewer_report\": reviewer_report,\n",
    "            \"documentation\": documentation\n",
    "        }\n",
    "    except Exception as e:\n",
    "        print(f\"Error in review process: {str(e)}\")\n",
    "        return {\n",
    "            \"error\": str(e)\n",
    "        }"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f1e79f26-0b6d-4677-9136-c8ee566e822e",
   "metadata": {},
   "source": [
    "### Testing the Multi-Agent System"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "ac2a9309-d328-493a-8597-d10bf641bbb9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting code review system...\n",
      "This will process code through three specialized agents in sequence:\n",
      "1. Code Checker: Checks for syntax errors and bugs\n",
      "2. Code Reviewer: Evaluates code quality and suggests improvements\n",
      "3. Documentation Generator: Creates HTML documentation with PDF export\n",
      "\n",
      "Processing sample code...\n",
      "Step 1: Running code checker...\n",
      "Code checker analysis complete!\n",
      "\n",
      "Step 2: Running code reviewer...\n",
      "Code review complete!\n",
      "\n",
      "Step 3: Running documentation generator...\n",
      "Documentation generation complete!\n",
      "\n",
      "Review process completed successfully!\n",
      "\n",
      "--- CODE CHECKER REPORT ---\n",
      "## Code Analysis\n",
      "\n",
      "### Syntax\n",
      "The code contains no syntax errors and is written in valid Python.\n",
      "\n",
      "### Functionality\n",
      "The function `calculate_average` is designed to compute the average of a list of numbers. Here is a step-by-step analysis:\n",
      "\n",
      "1. **Docstring**: The function is well-documented, including a clear explanation of its purpose, input arguments, return value, and potential exceptions. This improves code readability and maintainability.\n",
      "\n",
      "2. **Empty List Check**: \n",
      "   - The code correctly checks if the input list `numbers` is empty. If it is, a `ZeroDivisionError` is raised with an appropriate message. This prevents division by zero, which would occur when attempting to compute the average of an empty list.\n",
      "\n",
      "3. **Total and Average Calculation**:\n",
      "   - `total = sum(numbers)` efficiently computes the sum of the list.\n",
      "   - `return total / len(numbers)` computes and returns the average as a float.\n",
      "\n",
      "### Logic and Algorithm\n",
      "- The function accurately calculates the average by dividing the sum of the numbers by the count of numbers, using Python's built-in `sum()` and `len()` functions. \n",
      "- The logic is straightforward and efficient for small to moderately sized lists.\n",
      "\n",
      "### Edge Cases\n",
      "- **Empty List**: Correctly handled by raising an exception.\n",
      "- **Single Element List**: The function would return the value of the single element, which is mathematically correct.\n",
      "\n",
      "### Example Usage\n",
      "- The `if __name__ == \"__main__\":` block demonstrates correct usage.\n",
      "- It calculates the average of a sample list `[10, 15, 20, 25, 30]` and prints the result. This serves as a basic test of functionality.\n",
      "\n",
      "## Recommendations\n",
      "- Although the code handles exceptions well and is efficient for typical use cases, consider extending it to handle non-list inputs gracefully, using type checks if required.\n",
      "- You could add more sophisticated error handling or logging as needed for larger applications or where the input data source may be unreliable.\n",
      "\n",
      "Overall, the code is well-designed, efficient, and clearly written, demonstrating good practices in Python programming.\n",
      "\n",
      "--- CODE REVIEWER REPORT ---\n",
      "# Code Review\n",
      "\n",
      "The code provided is generally well-written, adhering to Python standards and demonstrating clarity in its functionality. Below are some observations and suggestions for improvement:\n",
      "\n",
      "## Code Quality and Readability\n",
      "\n",
      "1. **Documentation and Comments**:\n",
      "   - The docstring is clear and concise, describing the function's purpose, parameters, return value, and exceptions raised. This enhances readability and ease of maintenance.\n",
      "\n",
      "2. **Functionality**:\n",
      "   - The function effectively calculates the average of a list and handles edge cases like an empty list with a `ZeroDivisionError`.\n",
      "\n",
      "3. **Code Structure**:\n",
      "   - The structure is clean, with a clear separation between the function definition and example usage. This follows good practices in Python programming.\n",
      "\n",
      "## Suggestions for Improvement\n",
      "\n",
      "1. **Type Annotations**:\n",
      "   - Adding type annotations can help with readability and tools like linters or type checkers. Consider specifying the input type as `List[float]` (import `List` from `typing` if necessary).\n",
      "\n",
      "    ```python\n",
      "    from typing import List\n",
      "\n",
      "    def calculate_average(numbers: List[float]) -> float:\n",
      "    ```\n",
      "\n",
      "2. **Input Validation**:\n",
      "   - While your function handles empty lists, you might want to validate that the input is indeed a list and contains numeric values. This can prevent unexpected errors when the input is not well-formed.\n",
      "\n",
      "    ```python\n",
      "    if not isinstance(numbers, list) or not numbers:\n",
      "        raise ValueError(\"Input must be a non-empty list of numbers\")\n",
      "\n",
      "    if not all(isinstance(n, (int, float)) for n in numbers):\n",
      "        raise ValueError(\"All elements in the list must be numeric\")\n",
      "    ```\n",
      "\n",
      "3. **Improved Error Handling**:\n",
      "   - Instead of `ZeroDivisionError`, raising a `ValueError` for an empty list may be more appropriate, as the primary issue is invalid input rather than a calculation error.\n",
      "\n",
      "4. **Example Usage**:\n",
      "   - Consider adding more example cases, including edge cases, to demonstrate robustness:\n",
      "\n",
      "    ```python\n",
      "    if __name__ == \"__main__\":\n",
      "        examples = [\n",
      "            [10, 15, 20, 25, 30],\n",
      "            [42],\n",
      "            []\n",
      "        ]\n",
      "\n",
      "        for example in examples:\n",
      "            try:\n",
      "                avg = calculate_average(example)\n",
      "                print(f\"The average of {example} is: {avg}\")\n",
      "            except ValueError as e:\n",
      "                print(f\"Error: {e}\")\n",
      "    ```\n",
      "\n",
      "5. **Error Messages**:\n",
      "   - Make error messages more descriptive to aid debugging and readability for users.\n",
      "\n",
      "## Conclusion\n",
      "\n",
      "The code is largely efficient and adheres to best practices. By integrating type annotations and more comprehensive input validation, you can further enhance the robustness and clarity of your code.\n",
      "\n",
      "--- DOCUMENTATION TEXT ---\n",
      "Did not receive HTML. Here's the documentation text:\n",
      "Here is the documentation for the `calculate_average` function:\n",
      "\n",
      "[View Documentation](sandbox:/mnt/data/output.html)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'checker_report': '## Code Analysis\\n\\n### Syntax\\nThe code contains no syntax errors and is written in valid Python.\\n\\n### Functionality\\nThe function `calculate_average` is designed to compute the average of a list of numbers. Here is a step-by-step analysis:\\n\\n1. **Docstring**: The function is well-documented, including a clear explanation of its purpose, input arguments, return value, and potential exceptions. This improves code readability and maintainability.\\n\\n2. **Empty List Check**: \\n   - The code correctly checks if the input list `numbers` is empty. If it is, a `ZeroDivisionError` is raised with an appropriate message. This prevents division by zero, which would occur when attempting to compute the average of an empty list.\\n\\n3. **Total and Average Calculation**:\\n   - `total = sum(numbers)` efficiently computes the sum of the list.\\n   - `return total / len(numbers)` computes and returns the average as a float.\\n\\n### Logic and Algorithm\\n- The function accurately calculates the average by dividing the sum of the numbers by the count of numbers, using Python\\'s built-in `sum()` and `len()` functions. \\n- The logic is straightforward and efficient for small to moderately sized lists.\\n\\n### Edge Cases\\n- **Empty List**: Correctly handled by raising an exception.\\n- **Single Element List**: The function would return the value of the single element, which is mathematically correct.\\n\\n### Example Usage\\n- The `if __name__ == \"__main__\":` block demonstrates correct usage.\\n- It calculates the average of a sample list `[10, 15, 20, 25, 30]` and prints the result. This serves as a basic test of functionality.\\n\\n## Recommendations\\n- Although the code handles exceptions well and is efficient for typical use cases, consider extending it to handle non-list inputs gracefully, using type checks if required.\\n- You could add more sophisticated error handling or logging as needed for larger applications or where the input data source may be unreliable.\\n\\nOverall, the code is well-designed, efficient, and clearly written, demonstrating good practices in Python programming.',\n",
       " 'reviewer_report': '# Code Review\\n\\nThe code provided is generally well-written, adhering to Python standards and demonstrating clarity in its functionality. Below are some observations and suggestions for improvement:\\n\\n## Code Quality and Readability\\n\\n1. **Documentation and Comments**:\\n   - The docstring is clear and concise, describing the function\\'s purpose, parameters, return value, and exceptions raised. This enhances readability and ease of maintenance.\\n\\n2. **Functionality**:\\n   - The function effectively calculates the average of a list and handles edge cases like an empty list with a `ZeroDivisionError`.\\n\\n3. **Code Structure**:\\n   - The structure is clean, with a clear separation between the function definition and example usage. This follows good practices in Python programming.\\n\\n## Suggestions for Improvement\\n\\n1. **Type Annotations**:\\n   - Adding type annotations can help with readability and tools like linters or type checkers. Consider specifying the input type as `List[float]` (import `List` from `typing` if necessary).\\n\\n    ```python\\n    from typing import List\\n\\n    def calculate_average(numbers: List[float]) -> float:\\n    ```\\n\\n2. **Input Validation**:\\n   - While your function handles empty lists, you might want to validate that the input is indeed a list and contains numeric values. This can prevent unexpected errors when the input is not well-formed.\\n\\n    ```python\\n    if not isinstance(numbers, list) or not numbers:\\n        raise ValueError(\"Input must be a non-empty list of numbers\")\\n\\n    if not all(isinstance(n, (int, float)) for n in numbers):\\n        raise ValueError(\"All elements in the list must be numeric\")\\n    ```\\n\\n3. **Improved Error Handling**:\\n   - Instead of `ZeroDivisionError`, raising a `ValueError` for an empty list may be more appropriate, as the primary issue is invalid input rather than a calculation error.\\n\\n4. **Example Usage**:\\n   - Consider adding more example cases, including edge cases, to demonstrate robustness:\\n\\n    ```python\\n    if __name__ == \"__main__\":\\n        examples = [\\n            [10, 15, 20, 25, 30],\\n            [42],\\n            []\\n        ]\\n\\n        for example in examples:\\n            try:\\n                avg = calculate_average(example)\\n                print(f\"The average of {example} is: {avg}\")\\n            except ValueError as e:\\n                print(f\"Error: {e}\")\\n    ```\\n\\n5. **Error Messages**:\\n   - Make error messages more descriptive to aid debugging and readability for users.\\n\\n## Conclusion\\n\\nThe code is largely efficient and adheres to best practices. By integrating type annotations and more comprehensive input validation, you can further enhance the robustness and clarity of your code.',\n",
       " 'documentation': 'Here is the documentation for the `calculate_average` function:\\n\\n[View Documentation](sandbox:/mnt/data/output.html)'}"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Sample code to review\n",
    "sample_code = \"\"\"\n",
    "def calculate_average(numbers):\n",
    "    \\\"\\\"\\\"\n",
    "    Calculate the average of a list of numbers.\n",
    "    \n",
    "    Args:\n",
    "        numbers (list): A list of numeric values\n",
    "        \n",
    "    Returns:\n",
    "        float: The average of all values\n",
    "        \n",
    "    Raises:\n",
    "        ZeroDivisionError: If the list is empty\n",
    "    \\\"\\\"\\\"\n",
    "    if not numbers:\n",
    "        raise ZeroDivisionError(\"Cannot calculate average of empty list\")\n",
    "        \n",
    "    total = sum(numbers)\n",
    "    return total / len(numbers)\n",
    "\n",
    "# Example usage\n",
    "if __name__ == \"__main__\":\n",
    "    data = [10, 15, 20, 25, 30]\n",
    "    avg = calculate_average(data)\n",
    "    print(f\"The average is: {avg}\")\n",
    "\"\"\"\n",
    "\n",
    "async def run_code_review_system():\n",
    "    print(\"Starting code review system...\")\n",
    "    print(\"This will process code through three specialized agents in sequence:\")\n",
    "    print(\"1. Code Checker: Checks for syntax errors and bugs\")\n",
    "    print(\"2. Code Reviewer: Evaluates code quality and suggests improvements\")\n",
    "    print(\"3. Documentation Generator: Creates HTML documentation with PDF export\")\n",
    "    print(\"\\nProcessing sample code...\")\n",
    "    \n",
    "    results = await run_complete_review(sample_code)\n",
    "    \n",
    "    if \"error\" in results:\n",
    "        print(f\"\\nError occurred: {results['error']}\")\n",
    "        return results\n",
    "    \n",
    "    print(\"\\nReview process completed successfully!\")\n",
    "    \n",
    "    print(\"\\n--- CODE CHECKER REPORT ---\")\n",
    "    print(results[\"checker_report\"])\n",
    "    \n",
    "    print(\"\\n--- CODE REVIEWER REPORT ---\")\n",
    "    print(results[\"reviewer_report\"])\n",
    "    \n",
    "    documentation = results[\"documentation\"]\n",
    "    if documentation.startswith(\"<!DOCTYPE html>\") or \"<html>\" in documentation:\n",
    "        print(\"\\n--- DOCUMENTATION GENERATED ---\")\n",
    "        print(\"HTML documentation generated successfully!\")\n",
    "        \n",
    "        html_file = \"code_documentation.html\"\n",
    "        with open(html_file, \"w\") as f:\n",
    "            f.write(documentation)\n",
    "        \n",
    "        print(f\"\\nHTML documentation saved to {html_file}\")\n",
    "        print(\"You can open this file in your browser to view the documentation\")\n",
    "        print(\"and use the 'Export as PDF' button\")\n",
    "    else:\n",
    "        print(\"\\n--- DOCUMENTATION TEXT ---\")\n",
    "        print(\"Did not receive HTML. Here's the documentation text:\")\n",
    "        print(documentation)\n",
    "    \n",
    "    return results\n",
    "await run_code_review_system()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ae4279bf-bccf-4045-9675-93e22765c8f7",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d88edd79-fbdb-4e66-8bf8-b0d20a208c26",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8826a9a3-cf97-4100-8a7f-7a0a3da4a414",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "143aa2ba-56b3-4de0-9422-e6046e5f648e",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f655fc12-ef92-4f05-a227-ecd720b01315",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "602c1669-4172-4866-84de-d588969c11f7",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b28c71ec-69eb-40ea-8f49-4907ffc65569",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6d24e53b-72e7-4306-a631-973a1e531d66",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "75f3befc-ce8f-462c-abf9-dd1e52a84265",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f9babe33-24a0-4056-a64b-98ed1cf25f26",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8198395d-f94c-4973-b544-41c29b5f9ed5",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3530e09c-1fd3-423d-b9fa-6823d1a9e962",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1a1ba34a-5bb4-447c-834a-6385b1c704df",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3c918c4a-002d-40a4-9d49-e78c03bcf033",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a8612c47-8a38-4435-a673-54ad5a4ce4d3",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "dfe74288-0860-49b2-bb3c-484c07778c63",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4d745871-6344-4afd-9588-56d6d03baf73",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bedadbf2-7812-4021-b98c-c8ed1e1b719e",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "97c01909-8f45-4f30-9151-983d3afcb24b",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "99b8d206-78c2-4aea-82d0-8a0b6f567833",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "24bcdb61-db90-4f4f-a841-414ace645a64",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9e09d638-028a-43bc-b172-8f16b19d9d31",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "96698d0c-efb1-4131-ab31-5b776fa68788",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2d5e7d9d-3fe2-43b4-8134-ca8fd6060812",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b75dd181-1c20-4d6a-8e4e-b39195a73ed5",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "34264601-9328-4975-8e5c-888906e2b851",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "558406ac-9a68-4ef8-9962-d8a81a6ef1d9",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f2370558-7edd-4315-85e3-70358d937057",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8ac36b9b-a69b-4600-a0e2-82b03b45923f",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:base]",
   "language": "python",
   "name": "conda-base-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}