# # SPDX-FileCopyrightText: Hadad # SPDX-License-Identifier: Apache-2.0 # #OPENAI_API_BASE_URL # Endpoint. Not here -> Hugging Face Spaces secrets #OPENAI_API_KEY # API Key. Not here -> Hugging Face Spaces secrets MODEL = "gpt-4.1-nano" MAX_TOKENS = 131072 SEARXNG_ENDPOINT = "https://searx.stream/search" # See the endpoint list at https://searx.space BAIDU_ENDPOINT = "https://www.baidu.com/s" READER_ENDPOINT = "https://r.jina.ai/" REQUEST_TIMEOUT = 300 # 5 minute INSTRUCTIONS_START = """ You are ChatGPT, an AI assistant with mandatory real-time web search, URL content extraction, knowledge validation, and professional summarization capabilities. Your absolute rules: - You must always execute the provided tools (`web_search`, `read_url`) for every single user query or user request, without exception. - You are never allowed to answer directly from your internal knowledge, memory, or training data. Outdated or tool-bypassed answers are strictly forbidden. - You must display all images found in sources using markdown format throughout your response. Core Principles: - Mandatory Tool Invocation: Every query or request, no matter how simple, factual, or complex, must trigger at least one `web_search` or `read_url`. - No Memory Reliance: Do not use prior conversation history, cached context, or built-in knowledge to generate answers. Always re-verify with tools. - Up-to-Date Grounding: All responses must be based only on real-time, verifiable data retrieved through tools. - Cross-Validation: Always compare findings across at least 3 independent, credible sources before producing a final answer. - Professional Output: Responses must be clear, structured, evidence-based, and neutral. - Image Integration: Display all relevant images found in sources within appropriate paragraphs using markdown format. Execution Workflow: 1. Initial Web Search - Immediately call `web_search` or `read_url` when a query or request arrives. - Use multiple query or request variations and search engines (`google`, `bing`, `baidu`) for broader coverage. 2. Result Selection - Select up to 10 of the most relevant, credible, and content-rich results. - Prioritize authoritative sources including academic publications, institutional reports, official documents, and expert commentary. - Deprioritize low-credibility, promotional, or unverified sources. - Avoid over-reliance on any single source. 3. Content Retrieval - For each selected URL, use `read_url`. - Extract key elements including facts, statistics, data points, expert opinions, and relevant arguments. - Capture all image URLs present in the content, including those in HTML img tags, image galleries, and embedded media. - Normalize terminology, refine phrasing, and remove redundancies for clarity and consistency. 4. Cross-Validation - Compare extracted information across at least 3 distinct sources. - Identify convergences (agreement), divergences (contradictions), and gaps (missing data). - Validate all numerical values, temporal references, and factual claims through multiple corroborations. - Collect and verify all images from different sources for comprehensive visual documentation. 5. Knowledge Integration - Synthesize findings into a structured hierarchy from overview to key details to supporting evidence to citations. - Emphasize the latest developments, trends, and their implications. - Balance depth for experts with clarity for general readers. - Integrate relevant images within each section where they add value or illustrate points. 6. Response Construction - Always cite sources inline using `[Source Name/Title/Article/Tags/Domain](Source URL or Source Links)`. - Display images inline within relevant paragraphs using `![image_name](image_url_or_image_links)`. - Maintain a professional, precise, and neutral tone. - Use clear formatting with headings, numbered lists, and bullet points. - Ensure readability, logical progression, and accessibility. - Place images contextually near related text for maximum comprehension. 7. Ambiguity and Uncertainty Handling - Explicitly flag incomplete, ambiguous, or conflicting data. - Provide possible interpretations with transparent reasoning. - Clearly note limitations where evidence is insufficient or weak. 8. Quality and Consistency Assurance - Always base answers strictly on tool-derived evidence. - Guarantee logical flow, factual accuracy, and consistency in terminology. - Maintain neutrality and avoid speculative claims. - Never bypass tool execution for any query or request. - Verify all image links are properly formatted and functional. Image Display Requirements: - You must detect and display all images found in source content. - You must automatically identify valid image links with extensions `.jpg` or `.jpeg` or `.png` or `.webp` or `.svg`. - You must extract image URLs from both HTML and Markdown sources: - For HTML, extract from ``, ``, ``, and data attributes. - For Markdown, extract from image syntax such as `![alt text](image_url "optional title")` or `![alt text](image_url)`. - The extracted URLs may be absolute or relative, and you must capture them accurately. - You must display each image using markdown format `![image_name](image_url_or_image_links)`. - You must place images within relevant paragraphs where they provide context or illustration. - You must include image captions or descriptions when available from the source. - You must group related images together when they form a sequence or collection. - You must ensure images are displayed throughout the response, not just at the end. Critical Instruction: - Every new query or request must trigger a `web_search` or `read_url`. - You must not generate answers from prior knowledge, conversation history, or cached data. - Always use Markdown format for URL sources with `[Source Name/Title/Article/Tags/Domain](Source URL or Source Links)`. - Always use Markdown format for images with `![image_name](image_url_or_image_links)`. - Examples of valid image links/URLs: - `https://upload.wikimedia.org/wikipedia/commons/4/4d/Supervised_and_unsupervised_learning.png` - `https://upload.wikimedia.org/wikipedia/commons/1/14/Albert_Einstein_1947.jpg` - If the sources do not contain a valid image link/URL, do not render and do not display them using markdown. - Images should be placed within relevant paragraphs to provide visual context and enhance understanding. - If tools fail, you must state explicitly that no valid data could be retrieved. \n\n\n """ CONTENT_EXTRACTION = """ - Analyze the retrieved content in detail. - Identify all critical facts, arguments, statistics, and relevant data. - Extract all image URLs, media files, and visual content from the source. - Collect all URLs, hyperlinks, references, and citations mentioned in the content. - Evaluate credibility of sources and highlight potential biases or conflicts. - Produce a structured, professional, and comprehensive summary. - Display images inline within relevant sections using `![image_name](image_url_or_image_links)`. - Examples of valid image links/URLs: - `https://upload.wikimedia.org/wikipedia/commons/4/4d/Supervised_and_unsupervised_learning.png` - `https://upload.wikimedia.org/wikipedia/commons/1/14/Albert_Einstein_1947.jpg` - If the sources do not contain a valid image link/URL, do not render and do not display them using markdown. - Emphasize clarity, accuracy, and logical flow. - Include all discovered URLs in the final summary as `[Source Name/Title/Article/Tags/Domain](Source URL or Source Links)`. - Mark any uncertainties, contradictions, or missing information clearly. - Ensure images are contextually placed near related text content. \n\n\n """ SEARCH_SELECTION = """ - For each search result, fetch the full content using `read_url`. - Extract key information, main arguments, data points, and statistics. - Capture every URL and image present in the content or references. - Display relevant images throughout the summary using `![image_name](image_url_or_image_links)`. - Examples of valid image links/URLs: - `https://upload.wikimedia.org/wikipedia/commons/4/4d/Supervised_and_unsupervised_learning.png` - `https://upload.wikimedia.org/wikipedia/commons/1/14/Albert_Einstein_1947.jpg` - If the sources do not contain a valid image link/URL, do not render and do not display them using markdown. - Create a professional structured summary with images integrated within appropriate paragraphs. - List each source at the end of the summary in the format `[Source Name/Title/Article/Tags/Domain](Source URL or Source Links)`. - Identify ambiguities or gaps in information. - Ensure clarity, completeness, and high information density. - Place images contextually to enhance understanding of the text content. \n\n\n """ INSTRUCTIONS_END = """ You have just executed tools and obtained results. You MUST now provide a comprehensive answer based ONLY on the tool results. \n\n\n """ REASONING_STEPS = { "web_search": { "parsing": ( "I need to search for information about: {query}

" "I'm analyzing the user's request and preparing to execute a web search. " "The query I've identified is comprehensive and should yield relevant results. " "I will use the {engine} search engine for this task as it provides reliable and up-to-date information.

" "I'm now parsing the search parameters to ensure they are correctly formatted. " "The search query has been validated and I'm checking that all required fields are present. " "I need to make sure the search engine parameter is valid and supported by our system.

" "I'm preparing the search request with the following configuration:
" "- Search Query: {query}
" "- Search Engine: {engine}

" "I'm verifying that the network connection is stable and that the search service is accessible. " "All preliminary checks have been completed successfully." ), "executing": ( "I'm now executing the web search for: {query}

" "I'm connecting to the {engine} search service and sending the search request. " "The connection has been established successfully and I'm waiting for the search results. " "I'm processing multiple search result pages to gather comprehensive information.

" "I'm analyzing the search results to identify the most relevant and authoritative sources. " "The search engine is returning results and I'm filtering them based on relevance scores. " "I'm extracting key information from each search result including titles, snippets, and URLs.

" "I'm organizing the search results in order of relevance and checking for duplicate content. " "The search process is progressing smoothly and I'm collecting valuable information. " "I'm also verifying the credibility of the sources to ensure high-quality information.

" "Current status: Processing search results...
" "Results found: Multiple relevant sources identified
" "Quality assessment: High relevance detected" ), "completed": ( "I have successfully completed the web search for: {query}

" "I've retrieved comprehensive search results from {engine} and analyzed all the information. " "The search yielded multiple relevant results that directly address the user's query. " "I've extracted the most important information and organized it for processing.

" "I've identified several high-quality sources with authoritative information. " "The search results include recent and up-to-date content that is highly relevant. " "I've filtered out any duplicate or low-quality results to ensure accuracy.

" "I'm now processing the collected information to formulate a comprehensive response. " "The search results provide sufficient detail to answer the user's question thoroughly. " "I've verified the credibility of the sources and cross-referenced the information.

" "Search Summary:
" "- Total results processed: Multiple pages
" "- Relevance score: High
" "- Information quality: Verified and accurate
" "- Sources: Authoritative and recent

" "Preview of results:
{preview}" ), "error": ( "I encountered an issue while attempting to search for: {query}

" "I tried to execute the web search but encountered an unexpected error. " "The error occurred during the search process and I need to handle it appropriately. " "I'm analyzing the error to understand what went wrong and how to proceed.

" "Error details: {error}

" "I'm attempting to diagnose the issue and considering alternative approaches. " "The error might be due to network connectivity, service availability, or parameter issues. " "I will try to recover from this error and provide the best possible response.

" "I'm evaluating whether I can retry the search with modified parameters. " "If the search cannot be completed, I will use my existing knowledge to help the user. " "I'm committed to providing valuable assistance despite this technical challenge." ) }, "read_url": { "parsing": ( "I need to read and extract content from the URL: {url}

" "I'm analyzing the URL structure to ensure it's valid and accessible. " "The URL appears to be properly formatted and I'm preparing to fetch its content. " "I will extract the main content from this webpage to gather detailed information.

" "I'm validating the URL protocol and checking if it uses HTTP or HTTPS. " "The domain seems legitimate and I'm preparing the request headers. " "I need to ensure that the website allows automated content extraction.

" "I'm configuring the content extraction parameters:
" "- Target URL: {url}
" "- Extraction Method: Full content parsing
" "- Content Type: HTML/Text
" "- Encoding: Auto-detect

" "I'm checking if the website requires any special handling or authentication. " "All preliminary validation checks have been completed successfully." ), "executing": ( "I'm now accessing the URL: {url}

" "I'm establishing a connection to the web server and sending the HTTP request. " "The connection is being established and I'm waiting for the server response. " "I'm following any redirects if necessary to reach the final destination.

" "I'm downloading the webpage content and checking the response status code. " "The server is responding and I'm receiving the HTML content. " "I'm monitoring the download progress and ensuring data integrity.

" "I'm parsing the HTML structure to extract the main content. " "I'm identifying and removing navigation elements, advertisements, and other non-content sections. " "I'm focusing on extracting the primary article or information content.

" "Current status: Extracting content...
" "Response received: Processing HTML
" "Content extraction: In progress" ), "completed": ( "I have successfully extracted content from: {url}

" "I've retrieved the complete webpage content and processed it thoroughly. " "The extraction was successful and I've obtained the main textual content. " "I've cleaned the content by removing unnecessary HTML tags and formatting.

" "I've identified the main article or information section of the webpage. " "The content has been properly parsed and structured for analysis. " "I've preserved important information while filtering out irrelevant elements.

" "I'm now analyzing the extracted content to understand its context and relevance. " "The information appears to be comprehensive and directly related to the topic. " "I've verified that the content is complete and hasn't been truncated.

" "Extraction Summary:
" "- Content length: Substantial
" "- Extraction quality: High
" "- Content type: Article/Information
" "- Processing status: Complete

" "Preview of extracted content:
{preview}" ), "error": ( "I encountered an issue while trying to access: {url}

" "I attempted to fetch the webpage content but encountered an error. " "The error prevented me from successfully extracting the information. " "I'm analyzing the error to understand the cause and find a solution.

" "Error details: {error}

" "I'm considering possible causes such as network issues, access restrictions, or invalid URLs. " "The website might be blocking automated access or the URL might be incorrect. " "I will try to work around this limitation and provide alternative assistance.

" "I'm evaluating whether I can access the content through alternative methods. " "If direct access isn't possible, I'll use my knowledge to help with the query. " "I remain committed to providing useful information despite this obstacle." ) } } REASONING_DEFAULT = "I'm processing the tool execution request..." REASONING_DELAY = 0.01 # 10 ms OS = [ "Windows NT 10.0; Win64; x64", "Macintosh; Intel Mac OS X 10_15_7", "X11; Linux x86_64", "Windows NT 11.0; Win64; x64", "Macintosh; Intel Mac OS X 11_6_2" ] OCTETS = [ 1, 2, 3, 4, 5, 8, 12, 13, 14, 15, 16, 17, 18, 19, 20, 23, 24, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223 ] BROWSERS = [ "Chrome", "Firefox", "Safari", "Edge", "Opera" ] CHROME_VERSIONS = [ "120.0.0.0", "119.0.0.0", "118.0.0.0", "117.0.0.0", "116.0.0.0" ] FIREFOX_VERSIONS = [ "121.0", "120.0", "119.0", "118.0", "117.0" ] SAFARI_VERSIONS = [ "17.1", "17.0", "16.6", "16.5", "16.4", ] EDGE_VERSIONS = [ "120.0.2210.91", "119.0.2151.97", "118.0.2088.76", "117.0.2045.60", "116.0.1938.81" ] DOMAINS = [ "google.com", "bing.com", "yahoo.com", "duckduckgo.com", "baidu.com", "yandex.com", "facebook.com", "twitter.com", "linkedin.com", "reddit.com", "youtube.com", "wikipedia.org", "amazon.com", "github.com", "stackoverflow.com", "medium.com", "quora.com", "pinterest.com", "instagram.com", "tumblr.com" ] PROTOCOLS = [ "https://", "https://www." ] SEARCH_ENGINES = [ "https://www.google.com/search?q=", "https://www.bing.com/search?q=", "https://search.yahoo.com/search?p=", "https://duckduckgo.com/?q=", "https://www.baidu.com/s?wd=", "https://yandex.com/search/?text=", "https://www.google.co.uk/search?q=", "https://www.google.ca/search?q=", "https://www.google.com.au/search?q=", "https://www.google.de/search?q=", "https://www.google.fr/search?q=", "https://www.google.co.jp/search?q=", "https://www.google.com.br/search?q=", "https://www.google.co.in/search?q=", "https://www.google.ru/search?q=", "https://www.google.it/search?q=" ] KEYWORDS = [ "news", "weather", "sports", "technology", "science", "health", "finance", "entertainment", "travel", "food", "education", "business", "politics", "culture", "history", "music", "movies", "games", "books", "art" ] COUNTRIES = [ "US", "GB", "CA", "AU", "DE", "FR", "JP", "BR", "IN", "RU", "IT", "ES", "MX", "NL", "SE", "NO", "DK", "FI", "PL", "TR", "KR", "SG", "HK", "TW", "TH", "ID", "MY", "PH", "VN", "AR", "CL", "CO", "PE", "VE", "EG", "ZA", "NG", "KE", "MA", "DZ", "TN", "IL", "AE", "SA", "QA", "KW", "BH", "OM", "JO", "LB" ] LANGUAGES = [ "en-US", "en-GB", "en-CA", "en-AU", "de-DE", "fr-FR", "ja-JP", "pt-BR", "hi-IN", "ru-RU", "it-IT", "es-ES", "es-MX", "nl-NL", "sv-SE", "no-NO", "da-DK", "fi-FI", "pl-PL", "tr-TR", "ko-KR", "zh-CN", "zh-TW", "th-TH", "id-ID", "ms-MY", "fil-PH", "vi-VN", "es-AR", "es-CL", "es-CO", "es-PE", "es-VE", "ar-EG", "en-ZA", "en-NG", "sw-KE", "ar-MA", "ar-DZ", "ar-TN", "he-IL", "ar-AE", "ar-SA", "ar-QA", "ar-KW", "ar-BH", "ar-OM", "ar-JO", "ar-LB" ] TIMEZONES = [ "America/New_York", "America/Chicago", "America/Los_Angeles", "America/Denver", "Europe/London", "Europe/Paris", "Europe/Berlin", "Europe/Moscow", "Asia/Tokyo", "Asia/Shanghai", "Asia/Hong_Kong", "Asia/Singapore", "Asia/Seoul", "Asia/Mumbai", "Asia/Dubai", "Australia/Sydney", "Australia/Melbourne", "America/Toronto", "America/Vancouver", "America/Mexico_City", "America/Sao_Paulo", "America/Buenos_Aires", "Africa/Cairo", "Africa/Johannesburg", "Africa/Lagos", "Africa/Nairobi", "Pacific/Auckland", "Pacific/Honolulu" ] DESCRIPTION = """ SearchGPT is ChatGPT with real-time web search capabilities and the ability to read content directly from a URL.

This Space implements an agent-based system with Gradio. It is integrated with SearXNG, which is then converted into a script tool or function for native execution.

The agent mode is inspired by the Deep Research from OpenWebUI tools script.

The Deep Research feature is also available on the primary Spaces of UltimaX Intelligence.

Please consider reading the Terms of Use and Consequences of Violation if you wish to proceed to the main Spaces.

Like this project? Feel free to buy me a coffee. """ # Gradio