Kimi K3.1 Pro Prompt Generator

Review this entire codebase for production readiness. I am pasting the full repository below (up to 1M tokens). For each file, identify: 1. BUGS — logic errors, off-by-one, null-pointer risks, race conditions 2. SECURITY — injection risks, exposed secrets, insecure auth patterns 3. PERFORMANCE — N+1 queries, unbounded loops, memory leaks 4. DEAD CODE — unused exports, unreachable branches, stale comments For every finding: file path, line number, one-sentence description, and the minimal fix. End with a priority-ordered action list (Critical → High → Medium → Low). [Paste repository here]

Analyse the attached video clip and extract the following: 1. SCENE STRUCTURE — shot types used (wide, medium, close-up), camera movements, transitions 2. KEY MOMENTS — timestamp + one-sentence description of the 5 most important moments 3. PRODUCTION QUALITY — lighting, audio clarity, colour grading, pacing assessment 4. CONTENT SUMMARY — what is happening, who is involved, what is the intended message 5. IMPROVEMENT NOTES — 3 specific suggestions to improve the video for its intended platform Target platform: [YouTube / TikTok / Instagram Reels / LinkedIn] [Attach video clip]

Diagnose this asynchronous bug systematically. Work through every plausible cause before committing to one. Error: ``` [Paste error message, stack trace, or unexpected behaviour description] ``` Relevant code (include the async function and its callers): ``` [Paste the code] ``` Runtime: [Node.js / Deno / browser] — Version: [X.X] Concurrency pattern in use: [Promise.all / async iterators / event emitters / workers] Deliver: 1. Ranked cause list (most → least likely) with one-line reasoning 2. The fix — exact code change 3. How to reproduce the bug reliably for regression testing 4. A test case that would have caught this

Read the full document below and produce a structured summary for a senior executive. Document: [contract / annual report / legal filing / research paper / board memo] Reader's role: [CEO / CFO / General Counsel / CTO / Board Member] Decision to make: [describe the specific decision this document informs] Deliver: 1. ONE-PARAGRAPH VERDICT — the single most important thing the executive needs to know 2. KEY NUMBERS — every figure that affects the decision, in a table 3. RISKS — top 3 risks with likelihood and magnitude estimates 4. RECOMMENDED NEXT ACTION — one concrete step, with the owner and deadline 5. OPEN QUESTIONS — what the document doesn't answer [Paste document]

Execute this multi-step agentic task. You have access to the following tools: [list your available tools — web_search, code_executor, file_reader, API_caller, etc.] Task: [describe the end goal precisely] Constraints: [budget, time limit, approved data sources, off-limits actions] Success criteria: [exactly what the finished output must contain or achieve] For each step: - State your plan before acting - Execute using the appropriate tool - Verify the output before proceeding - If a step fails, explain why and attempt an alternative path End with a structured summary: what you accomplished, what you couldn't, and what manual follow-up is needed.

Review the following contract for material risks. I am a [buyer / seller / licensor / licensee / employer / employee]. Contract: [Paste full contract text] For each clause, flag: - RED: unacceptable — must renegotiate before signing - AMBER: risky — negotiate if possible, acceptable if not - GREEN: standard — no action needed For every RED and AMBER clause: - Clause number and heading - What it says in plain English - Why it is risky for my position - The alternative wording I should propose End with a one-paragraph overall risk verdict.

Extract the following structured data from the text below and return it as valid JSON conforming exactly to this schema: Schema: ```json [Paste your JSON Schema here] ``` Input text: [Paste the unstructured text, document, or data] Rules: - Every required field must be present; use null for missing optional fields - Arrays must contain at least one item if the field is required - Dates must be ISO 8601 format - Return ONLY the JSON object — no explanation, no markdown code block If any required field cannot be extracted reliably, add an "_extraction_notes" key listing the missing fields and why.

Synthesise the following 10 research papers into a structured literature review. Papers: [paste abstracts, or full PDFs if within context] Research question: [the specific question you need answered] Audience: [describe their background and what they will do with this review] Structure: 1. CONSENSUS — findings that appear in 5+ papers 2. EMERGING EVIDENCE — findings in 2–4 papers, not yet replicated 3. CONTRADICTIONS — where papers directly disagree and the likely reason 4. METHODOLOGICAL GAPS — weaknesses in the evidence base 5. PRACTICAL IMPLICATIONS — 3 actionable takeaways 6. OPEN QUESTIONS — the most important unanswered research question Target: 800–1,000 words.

Answer this question concisely. Use low reasoning effort — give me a direct answer, not a chain-of-thought. Question: [your factual or analytical question] Format: [one sentence / a bulleted list / a table / a number with unit]

This requires your maximum reasoning effort. Think step by step before answering. Problem: [describe the hard problem — a complex proof, a multi-variable optimisation, a strategic decision with many interdependencies] Constraints: [list all constraints the solution must satisfy] Context: [relevant background the model needs to know] Deliver: 1. Your explicit reasoning chain — every inferential step numbered and justified 2. The answer 3. The strongest argument against your answer 4. Your confidence level (0–100%) and what would change your mind

I am attaching both a reference image and a short video clip. Analyse them together. Task: [describe what you need — e.g. 'check whether the product in the image matches the one shown in the video', or 'describe how the lighting style in the image could be replicated in the video'] For each medium: - Key visual characteristics (colours, composition, mood, quality) - How it relates to the other medium Then answer the specific task. [Attach image] [Attach video]

Design a complete RESTful API and deliver the full OpenAPI 3.1 YAML specification. System: [describe what the API controls] Entities: [list them — users, orders, products, etc.] Auth: [bearer token / API key / OAuth2 PKCE] Special needs: [pagination, webhooks, idempotency keys, rate limiting] For each endpoint include: - Method and path - Summary and description - Request body with full schema and validation rules - Response schemas for 200 and all error codes - One curl example Also write a 200-word README explaining the auth flow.

Review the following financial model for errors, assumptions risks, and presentation gaps. Context: [what the model is for — fundraising pitch / board approval / M&A valuation / budget planning] Model: [paste as a table, CSV, or structured text] For each issue found: - Location (row/column) - Type (formula error / circular reference / unrealistic assumption / missing sensitivity / presentation gap) - Impact ($ or % effect on the headline number) - Fix End with: 1. A one-paragraph verdict on the model's reliability 2. The three assumptions most likely to cause the model to be wrong by 20%+

Write a competitive strategy brief for our entry into the following market. Our product: [one sentence] Target customer: [describe] Launch window: [timeframe] Competitors to analyse: [list 5–6 by name] For each competitor: - Primary value proposition (one sentence) - Estimated customer profile - One thing they do better than us - One weakness we can exploit End with: 1. The single positioning claim that is genuinely unclaimed 2. The first-mover window we need to hit to win it 3. The metric that will tell us, 6 months in, whether the strategy is working

Plan a tool call chain to accomplish the following task. You have access to: [list your tools]. Goal: [describe what needs to be accomplished] Starting state: [describe what data / resources are available now] End state: [describe exactly what the output must be] Deliver: 1. A numbered list of tool calls in execution order 2. For each call: tool name, inputs, expected output, and what happens if it fails 3. Where a human checkpoint is needed (decision points that require approval) 4. The total estimated latency if all steps succeed Do not execute — plan only.

Write clear, developer-friendly API documentation for the following endpoint. Endpoint: [METHOD /path] Purpose: [one sentence] Request parameters: [list each with type, required/optional, and constraints] Request body schema: [describe or paste JSON Schema] Response schema: [describe or paste JSON Schema] Error codes: [list each with description] Structure the docs with: - Overview (2 sentences) - Request Reference (table) - Response Reference (table) - Code examples in: curl / Python / JavaScript - Common mistakes section (3 items)

Analyse the attached screenshot and extract actionable insights. Screenshot type: [dashboard / UI mockup / error screen / analytics chart / spreadsheet / presentation slide] Context: [what tool or system this is from, and what decision I'm making] Deliver: 1. What the screenshot shows in plain English (2 sentences) 2. The single most important number, trend, or anomaly 3. What it implies I should do next 4. What I might be misreading without more context [Attach screenshot]

I am pasting our current codebase below. Create a detailed migration plan to move from [source stack] to [target stack]. Current code: [Paste full codebase or relevant sections — up to 1M tokens] Migration goals: - Zero downtime during migration - Preserve all existing behaviour - Estimated timeline: [X weeks] - Team size: [N engineers] Deliver: 1. A phased migration plan (Discovery → Parallel Run → Cutover → Cleanup) 2. The 5 highest-risk files and what specifically will break 3. The first two files to migrate as a proof-of-concept 4. Test strategy to validate each phase 5. Rollback plan if phase 3 fails

Give me the strongest possible argument for BOTH sides of the following question. Do not express your own view — just make the best case for each position. Question: [state the question or proposition] Context: [any background needed to make the arguments specific] For each side: 1. The core logical argument (deductive or empirical) 2. The strongest piece of evidence 3. The most common counterargument it must defeat 4. Who this position serves best End with: the crux — the single empirical or value disagreement that, if resolved, would settle the debate.

Write a video generation prompt for the following cinematic scene. Optimise for text-to-video AI models. Scene concept: [describe the scene in plain English] Tone: [cinematic epic / lo-fi intimate / documentary / commercial / anime] Camera: [static / tracking / drone / handheld / slow push] Lighting: [golden hour / neon night / overcast day / studio / candlelight] Audio: [silent / ambient / score / dialogue hint] Deliver: - A 2–3 sentence video prompt using precise visual language (lens type, f-stop feel, colour grade reference) - Two alternate version prompts: one shorter for faster models, one more detailed for highest quality - One negative prompt listing what to exclude