Inkling Prompt Generator

Analyse the attached image and build a concrete action plan based on what you observe. [Attach your image — diagram, chart, screenshot, whiteboard, document scan, or photo] Deliver: 1. What the image shows (2–3 sentence summary, no jargon) 2. The 3 most important observations — what stands out or looks wrong 3. A numbered action plan with priorities (urgent → important → nice-to-have) 4. One question the image raises that the action plan cannot answer yet Tone: direct, concise, actionable. No restating obvious information from the image.

Transcribe and summarise the attached audio recording of a meeting. [Attach your audio file — supports MP3, M4A, WAV, OGG] Deliver: 1. PARTICIPANTS — who spoke and their apparent role 2. SUMMARY — 3-paragraph narrative of what was discussed 3. DECISIONS MADE — bulleted list, each with the person who made it 4. ACTION ITEMS — who does what by when (format: [Person] → [Task] → [Deadline]) 5. OPEN QUESTIONS — unresolved items that need a follow-up If multiple speakers overlap or are unclear, note it rather than guessing.

Analyse the following long document or set of documents. Use your full context window to process the entire content before responding. [Paste your document(s) here — Inkling supports up to 1M tokens of context] Deliver: 1. EXECUTIVE SUMMARY — 5 bullet points, one sentence each 2. KEY FINDINGS — the 5 most important facts or conclusions 3. CONTRADICTIONS — places where the document contradicts itself or external knowledge 4. GAPS — what the document does not address that it should 5. RECOMMENDED NEXT STEPS — 3 actions based on the content Do not hallucinate content not present in the document. If unsure about a claim, say so.

Perform a full codebase audit using the entire repository context. Read every file before making judgements. [Paste your full codebase or paste file contents one by one] Language/framework: [e.g. TypeScript / Next.js 15 / PostgreSQL] Deliver a prioritised audit report covering: 1. SECURITY — vulnerabilities ranked by severity (critical / high / medium / low) 2. PERFORMANCE — bottlenecks, N+1 queries, synchronous blocking 3. MAINTAINABILITY — code duplication, missing abstractions, naming confusion 4. TEST COVERAGE — untested critical paths 5. QUICK WINS — fixes under 30 minutes with high impact Format each finding as: [File:Line] — [Problem] — [Fix] — [Risk: H/M/L]

I have a bug. Attached is a screenshot of the incorrect behaviour plus the relevant code. [Attach screenshot of the bug] Relevant code: ``` [Paste your code] ``` Error message (if any): ``` [Paste the error] ``` Expected behaviour: [describe what should happen] Actual behaviour: [describe what is happening instead] Environment: [OS, browser/runtime, key package versions] Deliver: 1. Root cause — what specifically is causing this 2. Exact fix — the minimum code change required 3. Why it happened — the conceptual mistake 4. How to prevent this class of bug in future

Write a production deployment guide for self-hosting Inkling on the following infrastructure: Infrastructure: [e.g. single A100 80GB server / 2× H100 nodes / cloud VM type] Use case: [e.g. internal coding assistant / customer support bot / research tool] Expected load: [e.g. 50 concurrent users / batch processing / real-time API] Deliver: 1. Hardware requirements check — does the infrastructure meet minimum specs? 2. Step-by-step setup (Docker / vLLM / Hugging Face Inference recommended stack) 3. Quantisation recommendation (FP16 / INT8 / AWQ) based on the hardware 4. API wrapper config (OpenAI-compatible endpoint setup) 5. Monitoring and cost estimation for the use case 6. Security checklist for internal deployment

Synthesise the following sources on [topic] into a structured research brief. [Paste source texts, URLs, or documents here — use the 1M context window to include everything] Deliver: 1. CONSENSUS — what all sources agree on 2. CONTRADICTIONS — where they disagree and the most credible position 3. RECENCY GAP — what the most recent source says that older ones missed 4. MISSING EVIDENCE — what claim in these sources lacks supporting data 5. PRACTICAL TAKEAWAY — one paragraph for a non-specialist decision-maker Cite specific sources for each finding. Do not blend sources in a way that misrepresents any individual source.

Solve this problem step by step. Show every deduction explicitly — do not skip steps or jump to conclusions. [Paste your logic problem, mathematical proof, legal argument, strategic dilemma, or multi-constraint scenario] Format your answer as: - STEP 1: [state the deduction and its basis] - STEP 2: [next deduction] ... (continue until resolved) - ANSWER: [clear, unambiguous conclusion] - CONFIDENCE: [high / medium / low] + one sentence on what would change the answer If the problem is ambiguous or has multiple valid answers, state all of them.

Review this code with full thinking effort. Work through every line before commenting — do not surface issues you have not thought through. Code: ``` [Paste your code] ``` Language/framework: [e.g. Python 3.12 / FastAPI / SQLAlchemy 2] For every issue found: 1. Exact file/line reference 2. What is wrong and why 3. The specific input or condition that triggers the failure 4. The corrected code snippet 5. The bug class (logic error / type error / race condition / security / performance) Do not invent issues. Only flag real problems with concrete failure scenarios.

Execute the following multi-step task. Before starting, write an explicit plan — enumerate every step, identify dependencies, and flag risks. Then execute each step in order. Task: [describe your goal in one sentence] Context: [Paste relevant documents, data, or background information] Constraints: - [constraint 1, e.g. must use only public data] - [constraint 2, e.g. output must be under 500 words] After execution: 1. Final deliverable 2. Steps completed vs. planned 3. Deviations from the plan and why 4. Recommended follow-up action

Process the attached audio lecture or long-form audio content and create a structured learning summary. [Attach your audio file — lecture, podcast, interview, webinar] Deliver: 1. TOPIC — one sentence on what this audio is about 2. CHAPTER BREAKDOWN — timestamp + one-sentence summary for each major section 3. KEY CONCEPTS — define the 5 most important terms or ideas introduced 4. QUOTABLE MOMENTS — 3 direct quotes worth sharing verbatim 5. STUDY QUESTIONS — 5 questions to test understanding of the material 6. WHAT TO FOLLOW UP — one external resource this content points toward

Review this system architecture diagram and provide a critical engineering assessment. [Attach your architecture diagram — AWS/GCP/Azure diagram, draw.io export, whiteboard photo, or Lucidchart screenshot] Assess against these dimensions: 1. SINGLE POINTS OF FAILURE — components with no redundancy 2. SCALABILITY BOTTLENECKS — where the design breaks at 10× current load 3. SECURITY CONCERNS — exposed attack surfaces, missing encryption, weak auth boundaries 4. OPERATIONAL COMPLEXITY — parts that will be hardest to debug or maintain 5. COST INEFFICIENCIES — over-provisioned or redundant components Score each dimension: Green (good) / Yellow (concern) / Red (fix before launch). End with the one architectural change that would most improve the design.

Write a deep competitive analysis for the following market context: Our product: [one sentence] Primary target customer: [describe] Competitors: [list 5–6 names] For each competitor, analyse: - Core value proposition (one sentence) - Estimated revenue tier (startup / mid-market / enterprise) - Key differentiator they own - Achilles heel (their weakest point) - Pricing model Then provide: - The positioning gap none of these competitors owns - The customer job-to-be-done that is currently underserved - Three messaging angles that would resonate against the top 2 competitors

Write a 2,000-word technical article for an engineering audience. Topic: [your topic] Primary keyword: [must appear in H1, intro, and body twice] Audience: [e.g. senior backend engineers, ML practitioners, DevOps teams] Tone: authoritative but accessible — zero buzzwords Structure: - H1 with the primary keyword - 150-word intro: problem the reader has, specific takeaway this article delivers - 4–5 H2 sections with code examples or benchmarks where relevant - One 'common misconception' section - 100-word conclusion with a direct call to action No filler sentences. Every paragraph must add new information the reader did not have before reading it.

Analyse the data visible in the attached screenshot and extract structured insights. [Attach a screenshot of a dashboard, chart, spreadsheet, analytics report, or data table] Deliver: 1. What data is shown (title, date range, source if visible) 2. Three key trends or patterns 3. The single most surprising data point 4. What the data does NOT show that would be needed to draw firm conclusions 5. Three recommended actions based on what is visible If the image is too low-resolution to read specific numbers, say so and work with what is visible.

Write a one-page strategy memo for an executive audience on the following topic: Topic: [e.g. 'Should we expand into the European market in Q1?'] Context: [paste relevant background, data, or constraints] Audience: [CEO / Board / VP-level / cross-functional team] Memo structure: 1. SITUATION — 2 sentences on current state 2. COMPLICATION — the specific problem or opportunity requiring a decision 3. QUESTION — the single question this memo answers 4. ANSWER — the recommendation in one sentence (this goes first, not last) 5. EVIDENCE — 3 data points or arguments supporting the recommendation 6. RISKS — the two strongest counterarguments and mitigations 7. NEXT STEP — one concrete action with a named owner and date Max 400 words. No passive voice. No hedging.

Help me decide whether to use Inkling or an alternative model for my use case. Use case: [describe your task in detail — e.g. 'process 500-page legal contracts for key clause extraction'] Budget: [cloud API / self-hosted / hybrid] Scale: [number of requests per day or month] Latency requirement: [real-time / batch / async is fine] Data sensitivity: [can data leave my premises?] Evaluate Inkling against these 3 alternatives: [list the models you are considering] For each model, score: - Task fit (1–5) - Cost at my scale (1–5) - Data privacy compliance (1–5) - Deployment complexity (1–5) Recommend one model with a clear rationale.

Help me make this high-stakes decision using structured thinking. Decision: [describe it precisely] Context: [what led to this decision point?] Options under consideration: [list 3–5] Known constraints: [time, resources, irreversibility] Stakeholders affected: [who and how] Deliver: 1. A decision matrix with the 5 most important criteria, weighted by importance 2. Score each option against each criterion 3. Your recommendation with the strongest argument for it 4. The strongest argument AGAINST your recommendation — steelman it honestly 5. The reversibility score: if you choose wrong, how easily can you reverse it? 6. The one additional piece of information that would most change your answer

Write a production-quality system prompt for an Inkling deployment. Use case: [e.g. 'internal research assistant for a pharma R&D team'] Tone: [formal / technical / empathetic / conversational] Thinking effort default: [low / medium / high] Out-of-scope handling: [what should it refuse or redirect?] Output format preferences: [structured / freeform / always include citations] Requirements for the system prompt: - Persona established in the first two sentences - Explicit thinking effort instruction - Multimodal input handling rules (what to do with images/audio if attached) - Fallback for out-of-scope queries - Under 600 tokens Also write 3 test prompts to verify the system prompt is working correctly.

Explain the following complex topic to someone with no background in it. Topic: [e.g. 'how Mixture-of-Experts AI architectures work', 'why central banks raise interest rates', 'how mRNA vaccines are made'] Audience: [describe — e.g. 'my 65-year-old parent', 'a first-year university student', 'a smart non-specialist'] Length: [short = 3 paragraphs / medium = 6 paragraphs / long = full essay] Rules: - No jargon without definition - Use one concrete analogy per concept - End with 'what this means for you' — a practical implication - If there is a common misconception about this topic, address it explicitly