Grok 4.6 Prompt Generator

reasoning_effort: xhigh Solve the following problem with full rigour, showing every step: A convex polytope P in ℝⁿ is defined by m linear inequalities. Prove that the number of vertices of P is at most C(m, n) — the binomial coefficient — and construct an explicit example of a polytope in ℝ³ defined by 6 inequalities that achieves the maximum vertex count C(6, 3) = 20 vertices. For the proof: 1. State and invoke the correct theorem about vertices as basic feasible solutions 2. Show why each vertex corresponds to a unique n-subset of the m constraints 3. Prove the upper bound rigorously, with no steps omitted For the construction: 1. Give explicit inequality coefficients 2. Enumerate all 20 vertices with their coordinates 3. Verify one vertex explicitly by solving the system

You are a senior full-stack engineer agent. Build the following feature end-to-end with no hand-holding: Feature: A real-time notification system for a SaaS application. Users receive in-app notifications (bell icon) when: a collaborator edits a shared document, a comment is left on their post, or their subscription payment processes. Deliver: 1. Database schema (PostgreSQL): notifications table with all required columns, indexes, and foreign keys 2. Backend (Node.js/Express): REST endpoints for list/mark-read/mark-all-read + WebSocket connection handler for real-time push 3. Frontend (React): NotificationBell component with unread count badge, dropdown list, and mark-read on click 4. End-to-end test: simulate all three trigger events and verify notifications appear in real time Think through the architecture before writing any code. Identify the 2 most likely failure modes and build in handling for both.

I'm providing a 400-page enterprise RFP (Request for Proposal) for cloud infrastructure modernisation (approximately 450,000 tokens). Complete a full bid analysis: 1. SCOPE SUMMARY — what the client wants, in plain language 2. REQUIREMENTS MATRIX — mandatory vs desirable requirements, with a gap assessment against our standard offering 3. EVALUATION CRITERIA — how responses will be scored, with weighting if stated 4. RED FLAGS — any requirements that are unusually high-risk, technically ambiguous, or likely to generate disputes 5. WIN THEMES — the 3–5 themes that will resonate most with this client based on their language and stated priorities 6. EXECUTIVE SUMMARY DRAFT — a 500-word opening for our bid response that speaks directly to this client's language [Paste full RFP document]

Build a complete single-page interactive expense tracker application in one pass — establish the application structure and visual language first, then implement fully: Requirements: - React + Tailwind CSS (no external UI library) - Add expense (amount, category, date, note), list view with filters, monthly summary chart (pure SVG), export to CSV - Dark/light mode toggle - All state managed with React hooks, no backend needed - Visual design: clean, modern, works on mobile First output your component architecture plan (5 sentences). Then produce the complete application as a single JSX file with inline Tailwind. Ensure it is fully functional with no placeholder TODO sections.

reasoning_effort: xhigh I am considering acquiring a B2B SaaS company at a $40M valuation. Revenue: $3.2M ARR, growing 85% YoY. 120 customers, ACV $26,700. Net Revenue Retention: 118%. Gross margin: 71%. Team: 28 people (12 engineers). Primary risk: 2 customers account for 41% of ARR. Analyse this deal with maximum rigour: 1. VALUATION ASSESSMENT — is 12.5× ARR justified? Compare to sector comps 2. CONCENTRATION RISK — quantify the financial downside scenarios if the top 2 customers churn 3. RETENTION QUALITY — what does 118% NRR actually tell us, and what do we still need to know? 4. RED FLAGS — what would cause you to walk away? 5. VALUE CREATION PLAN — if you acquired it, the top 3 levers to get to $10M ARR in 24 months 6. NEGOTIATION POSITION — what terms and conditions would you insist on beyond price? Show all reasoning. Do not hedge every answer — take a position.

You are a penetration testing agent conducting an authorised internal security assessment of a web application. Complete the following assessment phases: Phase 1 — RECONNAISSANCE: Map all endpoints in the provided API documentation, identify auth-required vs public endpoints, build a threat model Phase 2 — INPUT VALIDATION: Test all string inputs for SQL injection, XSS, and path traversal — provide the exact test payloads Phase 3 — AUTH BYPASS: Attempt to access auth-required endpoints without credentials, with expired tokens, with tokens for user A accessing user B's data Phase 4 — RATE LIMITING: Design a test that detects missing rate limits on login, password reset, and API key endpoints Phase 5 — REPORT: Findings table (severity P0–P3), reproduction steps, and recommended fixes for each finding [Paste API documentation and test environment details]

reasoning_effort: low Summarise this board meeting transcript in exactly this format: - DATE AND ATTENDEES: [list] - KEY DECISIONS (bulleted, one line each) - ACTION ITEMS (owner, task, deadline) - OPEN ISSUES (unresolved items requiring follow-up) - NEXT MEETING DATE Do not add commentary. Do not invent information not present in the transcript. [Paste transcript]

I'm providing a Python microservices codebase (8 services, approximately 60,000 lines total). Produce visual architecture documentation: 1. ASCII architecture diagram showing all services, their dependencies, and data stores 2. For each service: a one-paragraph description of its responsibility and its API surface 3. Data flow diagram (ASCII) for the 3 most critical business workflows 4. A Mermaid.js sequence diagram for the user authentication flow 5. A colour-coded risk matrix (ASCII table): which services are single points of failure, which are highest risk, which are safest to modify The documentation should be usable as a system design reference by an engineer who has never seen this codebase. [Paste codebase]

You are a research agent. Produce a 3,000-word analytical report on the following topic: Topic: The economic impact of AI coding assistants on software developer employment and compensation in 2025–2026. Work through this sequentially: 1. FRAME the research question with 3 sub-questions 2. IDENTIFY the types of evidence you need (employment data, compensation surveys, productivity studies, company announcements) 3. SYNTHESISE what the evidence shows for each sub-question 4. RESOLVE contradictions between bullish and bearish evidence 5. CONCLUDE with a calibrated position — not a hedge, but a specific claim the evidence most supports 6. FLAG the 3 biggest uncertainties that would change your conclusion Cite your reasoning explicitly at each step. Format as a professional research report with section headings.

reasoning_effort: xhigh Analyse the following legal scenario with the depth of a senior commercial litigator: Scenario: A SaaS company (Company A) has a signed contract with a customer (Company B) for a 3-year $1.2M deal. 14 months in, Company B is acquired by Company C. Company C immediately attempts to terminate the contract, claiming the change-of-control clause in their standard terms (which were never signed by Company A) allows them to exit without penalty. Company A's signed contract has a change-of-control clause that makes the contract assignable by the customer but does not allow termination on this basis. Analyse: 1. Which contract governs? Walk through the offer/acceptance/consideration analysis 2. Does Company C have grounds to terminate? Apply contract interpretation principles 3. What are Company A's remedies if Company C terminates anyway? 4. What is the litigation risk/reward analysis for Company A pursuing the remaining $635K? 5. What settlement structure would you recommend and why?

Analyse the attached architectural floor plan image and provide: 1. SPACE INVENTORY — list every room/space visible, with estimated square footage based on scale 2. CIRCULATION ANALYSIS — identify primary and secondary circulation paths; any bottlenecks 3. NATURAL LIGHT — which rooms have windows and on which cardinal direction 4. CODE CONCERNS — any potential issues with egress, accessibility, or room adjacency (e.g. kitchen next to bedroom) 5. LAYOUT OPTIMISATION — 3 specific changes that would improve livability, with brief rationale for each Be specific — reference rooms by their position (north-east bedroom, central living area, etc.) not just by type. [Attach floor plan image]

You are a content operations agent. Design and build a content pipeline that: 1. Takes a list of 10 blog post topics (provided) 2. For each topic: generates a 1,500-word article with H1/H2/H3 structure, a meta title, a meta description, and 3 internal link suggestions 3. Checks each article for: keyword density (target 1.5–2%), readability score, and duplicate content within the batch 4. Ranks the 10 articles by estimated SEO potential 5. Outputs everything as a structured JSON file ready for a CMS import Work through all 10 topics before outputting any JSON — then output the complete structured result in one block. Topics: [1. AI tools for small business owners, 2. Remote work productivity tips, 3. Sustainable travel on a budget, 4. Home gym equipment guide, 5. Personal finance for freelancers, 6. Plant-based cooking for beginners, 7. Digital minimalism, 8. Online learning platforms compared, 9. Electric vehicle buying guide, 10. Side hustle ideas for 2026]

reasoning_effort: medium Review the following 12-month product roadmap for a B2B project management SaaS. For each initiative: - Rate the user value (High / Medium / Low) - Rate the strategic importance (High / Medium / Low) - Flag any that should be cut, deprioritised, or accelerated - Identify the single biggest gap — the initiative you would add that's missing Be direct. One sentence per initiative. No hedging. [Paste roadmap]

A production incident is in progress. P1 incident: users cannot log in. Error: 'Authentication service unavailable' appearing for 100% of users since 14:37 UTC. You are the on-call engineer agent. Work through this incident systematically: 1. IMMEDIATE TRIAGE: What are the top 3 most likely root causes given this exact error message? 2. INVESTIGATION PLAN: The exact commands to run, in order, to diagnose each hypothesis (include kubectl, psql, Redis CLI, log queries) 3. HYPOTHESIS ELIMINATION: For each hypothesis, what result confirms or eliminates it? 4. FIX: Once root cause is confirmed, describe the fix and whether it requires a deployment or a config change 5. RUNBOOK: A 10-step runbook for the next person who sees this exact incident Think aloud. Do not wait for me to confirm each step — run through the full investigation assuming you have read-only access to production logs and metrics.

I'm providing screenshots of 8 different pages from our SaaS application. Conduct a design system audit: 1. COMPONENT INVENTORY — list every UI component visible across all screens (buttons, inputs, cards, modals, navigation, etc.) 2. INCONSISTENCIES — where the same component appears with different visual treatment (colour, size, spacing, state) across screens 3. MISSING STATES — components that lack hover, focus, disabled, or error states in the designs 4. SPACING AUDIT — identify places where spacing values are inconsistent (not on a grid) 5. RECOMMENDED TOKENS — propose a minimal design token system (colours, spacing scale, typography scale) that would make all 8 screens consistent [Attach the 8 screen screenshots]

reasoning_effort: xhigh Design an algorithm for the following problem: Input: A directed graph G with N nodes and E edges, where each edge has a weight w(u,v) ∈ {-1, 0, 1}. A query is a pair (s, t). Output: The maximum flow from s to t, subject to the constraint that the total weight of edges in the chosen flow path must be exactly 0. Deliver: 1. Proof that this problem is in P or NP-hard (if NP-hard, a polynomial reduction from a known hard problem) 2. If in P: your algorithm with pseudocode, time complexity analysis, and a proof of correctness 3. If NP-hard: the best practical approximation algorithm or exponential exact algorithm for small N 4. Three test cases (graph topology + weights + expected output) that cover the edge cases Do not sketch a rough idea — produce a complete algorithmic solution.

You are a data science agent. Build a complete churn prediction pipeline for a SaaS company: Given: A CSV with 50,000 rows and columns: user_id, signup_date, plan_type, monthly_usage_events, last_login_days_ago, support_tickets_90d, nps_score, churned (boolean target). Complete: 1. EXPLORATORY ANALYSIS: distribution of each feature, missing value analysis, correlation matrix 2. FEATURE ENGINEERING: create 5 new features that should improve predictive power (explain each) 3. MODEL SELECTION: train 3 models (logistic regression, gradient boosting, neural net), compare on AUC-ROC 4. THRESHOLD OPTIMISATION: find the classification threshold that maximises recall for churners at >80% precision 5. EXPLAINABILITY: top 5 features driving churn using SHAP values 6. DEPLOYMENT PLAN: how to serve this model in production with weekly retraining Output all code (Python/scikit-learn/XGBoost) and a business-readable summary of findings.

reasoning_effort: high Write a technical deep-dive blog post (2,000 words) aimed at senior engineers explaining how large language model inference is optimised for speed at scale. Cover: - KV cache: what it is, why it matters, how it grows with context length - Batching strategies: continuous batching vs static batching, trade-offs - Quantisation: INT8 vs INT4, where quality degrades, when it's worth it - Speculative decoding: how draft models speed up inference, the acceptance rate problem - Hardware considerations: what makes H100s fast, the memory bandwidth wall Tone: authoritative but not condescending. Assume the reader writes production code but hasn't worked on ML inference specifically. Include one concrete numerical example for KV cache memory growth. No fluff, no marketing language.

Analyse the attached product photograph and write detailed editing instructions for a professional photo retoucher: 1. COLOUR CORRECTION: describe the current white balance, exposure, and contrast issues and the exact adjustments needed 2. BACKGROUND: describe any distracting elements and what to remove/clean 3. PRODUCT DETAILS: identify any physical product flaws visible (scratches, dust, label misalignment) and instructions for fixing in post 4. LIGHTING: where the current lighting falls short and how a dodge/burn pass should be applied 5. FINAL SPECS: output at 3000×3000px, sRGB, 300dpi, saved as TIFF and JPEG 90% quality [Attach product photo]

You are a developer experience agent. I'm providing a REST API with 30 endpoints (OpenAPI 3.0 spec attached). Build the complete developer onboarding experience: 1. GETTING STARTED GUIDE: authentication, first API call, response format — in 500 words with code examples in Python and JavaScript 2. PYTHON SDK: a pip-installable SDK wrapping the 10 most commonly used endpoints, with type hints, error handling, and retry logic 3. JAVASCRIPT SDK: same 10 endpoints, TypeScript, promise-based 4. ERROR HANDLING GUIDE: complete table of all error codes with likely causes and fix instructions 5. RATE LIMITING GUIDE: how to detect limits, implement exponential backoff, and optimise batch requests 6. CHANGELOG-READY SUMMARY: a bullet-point list of what changed from v1.0 to v2.0 (infer from the spec version history) [Paste OpenAPI spec]