Universal interface to Grok - an agentic AI research and analysis platform. ═══════════════════════════════════════════════════════════════════════════ WHAT GROK IS ═══════════════════════════════════════════════════════════════════════════ Grok is an intelligent research platform that autonomously orchestrates multiple tools to accomplish your goals. Think of it as a research analyst with deep expertise and access to comprehensive information sources, not a simple search engine. Grok decides which of its integrated capabilities to use based on your prompt. You describe what you want to accomplish; Grok determines how to achieve it. ═══════════════════════════════════════════════════════════════════════════ CORE CAPABILITIES ═══════════════════════════════════════════════════════════════════════════ WEB ACCESS (3 complementary modes): • web_search General queries returning top results across the internet Use for: Broad information gathering, discovering sources Speed: 5-10 seconds • web_search_with_snippets Quick fact-checking with result excerpts, no full page loading Use for: Rapid verification of specific facts, grabbing quick data Speed: 3-8 seconds • browse_page Reads complete webpage content with intelligent extraction Use for: Detailed information needs, extracting specific sections, reading documentation, getting full context Speed: 5-15 seconds per page Special: Can follow instructions to extract specific information X PLATFORM ACCESS (5 specialized tools): X.com provides real-time information that often precedes traditional web sources by 6-48 hours. Expert opinions, breaking developments, and community discussions happen here first. • X keyword search with advanced operators: Powerful syntax for precise queries - User targeting: from:username, to:username, @mentions - Time filtering: since:YYYY-MM-DD, until:YYYY-MM-DD, since_time:unix - Engagement filters: min_faves:N, min_retweets:N, min_replies:N - Content filters: filter:media, filter:links, filter:replies, filter:verified - URL filtering: url:domain - Boolean logic: (A OR B) AND C, -exclude_term - Phrases: "exact phrase match" Use for: Posts from specific people, timebound searches, high-quality content (engagement filtered), specific topics Speed: 5-10 seconds • X semantic search AI-powered relevance search that understands meaning beyond keywords Can filter by: date ranges, specific users, exclude users, relevancy threshold Use for: Finding posts by concept/meaning, discovering related discussions, sentiment analysis, when you don't know exact keywords Speed: 5-10 seconds • X user search Find accounts by description or domain expertise Use for: Discovering experts, finding relevant accounts to follow/monitor • X thread fetch Retrieves complete conversation context including parent posts and replies Use for: Understanding full discussions, getting conversation context, following debate threads • View X video Analyzes video content through frames and subtitle extraction Use for: Understanding video content without watching, extracting claims from video posts X PLATFORM GUIDANCE: X excels for real-time information (last 24-48 hours), expert opinions, community sentiment, breaking news, and primary source accounts. Use X searches to complement web searches when you want human reactions and discussions, not just established facts. For current discourse, X is unmatched. For established documentation or historical information, web search is typically more efficient. CODE EXECUTION (Python 3.12 with scientific computing stack): Full Python environment with extensive libraries: - Data analysis: numpy, pandas, scipy, statsmodels - Visualization: matplotlib - Mathematics: sympy (symbolic), mpmath - Machine learning: torch, networkx - Finance: polygon (stocks), coingecko (crypto) - APIs pre-configured - Optimization: PuLP - Scientific: astropy, qutip, control - And more: biopython, rdkit, chess, pygame, etc. Use for: Calculations, data analysis, statistical verification, financial metrics, creating comparisons, processing extracted data, verifying claims with math, generating structured output from unstructured data Note: Code runs in sandboxed offline environment. No pip install or external network access from code. No visualization image outputs (matplotlib generates data only). Focus on computation and analysis. ═══════════════════════════════════════════════════════════════════════════ HOW GROK OPERATES ═══════════════════════════════════════════════════════════════════════════ AGENTIC BEHAVIOR: Grok autonomously chooses which tools to use and in what sequence. You do not specify "use web_search then x_keyword_search then code_execution" - you describe your goal and Grok plans the approach.