GPT-6.1 Terra Prompt Generator

Open [application name] on my desktop and complete the following task: Task: [describe exactly what you want done — e.g. 'Export all rows from the Customers tab of the spreadsheet as a CSV file named customers_export.csv to the Downloads folder'] Constraints: - Do not open any browser tab or application not required for this task - If a dialog or prompt appears that is ambiguous, stop and describe what you see rather than guessing - Confirm each major step before proceeding to the next When complete, describe: what you did, which files were affected, and the exact state of the application when you finished.

Extract all structured data from this screenshot. [attach screenshot] Output format: [table / JSON / bullet list / CSV] Fields to extract: [list the specific fields — e.g. name, date, amount, status] If a field is partially visible or cut off: include it with a note that it is truncated If text is ambiguous due to image quality: flag it with a confidence marker (HIGH / MEDIUM / LOW) Return the extracted data in the requested format, followed by a one-line summary of what the screenshot shows and any data quality issues I should be aware of.

Analyse this screenshot of a UI and write a complete bug report. [attach screenshot] Product context: [what product or page is this — e.g. 'our SaaS dashboard's billing page'] Expected behaviour: [what should this UI look like or do?] Browser/device context: [e.g. Chrome 130, macOS, 1440px viewport] Bug report format: 1. TITLE (one sentence, starts with the affected component) 2. SEVERITY (Critical / High / Medium / Low) with rationale 3. STEPS TO REPRODUCE (numbered, precise enough that a developer could follow them) 4. ACTUAL BEHAVIOUR (what the screenshot shows is happening) 5. EXPECTED BEHAVIOUR (what should happen instead) 6. LIKELY CAUSE (your hypothesis — CSS, state management, API response, etc.) 7. SUGGESTED FIX (the minimal change most likely to resolve it)

Summarise the following document at three levels of detail. Document: [paste or attach the document] Audience: [who will read this summary — exec / customer / developer / general?] Level 1 — EXECUTIVE (3 bullets, each under 20 words): the single most important fact from each major section Level 2 — OPERATIONAL (one paragraph, 100–150 words): what decision or action this document supports, who is responsible, and the key deadline or constraint Level 3 — DETAILED (structured outline): headers matching the document's own structure, one or two sentences per section capturing the essential content Flag any section where the source material is ambiguous, contradictory, or missing information a reader would need.

Write a Playwright automation script to fill and submit the following web form. Form URL: [the page URL] Form purpose: [what the form does — e.g. 'new user registration', 'contact inquiry'] Fields and values: [list each field label and the value to enter — e.g. 'First Name: John', 'Email: test@example.com'] Submit button: [describe the button text or selector] Success condition: [what appears on screen after successful submission] Script requirements: - Use Playwright with async/await - Add a 500ms wait after each field to avoid race conditions - Screenshot on success and on any error - Wrap in a try/catch that logs the failing field and the error message - Parameterise the field values so the script can be reused with different data

Write the following piece of content in clear, natural prose. Content type: [email / announcement / help article / product description / blog post intro / other] Audience: [who will read this — customers / employees / developers / general public?] Tone: [professional / friendly / direct / empathetic — pick one] Length: [target word count or range] Key information to include: - [fact or point 1] - [fact or point 2] - [fact or point 3] What the reader should do or know after reading: [the single call to action or takeaway] Write the full piece. No filler phrases ('In today's fast-paced world…'), no clichés, no bullet points unless the format explicitly calls for them. Sentences should be short enough to read at a glance.

Write an alt-text description and an extended image description for the following image. [attach image] Context: [where this image appears — e.g. 'hero image on our product homepage', 'thumbnail for a news article about climate policy'] Audience: [who might rely on this description — screen reader users, low-vision readers, SEO crawlers?] Output two descriptions: 1. ALT TEXT (under 125 characters): the essential information a sighted user gets at a glance — what is shown, not what the image is called 2. EXTENDED DESCRIPTION (2–4 sentences): spatial layout, key colours or visual contrasts, any text visible in the image, and the mood or context the image conveys Avoid starting with 'Image of' or 'Photo of'. Describe, don't label.

Draft a response to the following email. Original email: [paste the email you received] My intended response in plain terms: [summarise what you want to say in 1–3 bullet points — don't worry about wording] Tone: [formal / friendly / direct — pick one] Length: [short (under 100 words) / medium (100–200 words) / full reply (no limit)] Anything to avoid: [topics, commitments, or phrasing you don't want in the reply] Write the full email. Subject line if replying to a thread: Re: [original subject]. No placeholder text. If you need a piece of information to complete the reply that I haven't given you, ask for it rather than making something up.

Write a product description for the following item. Product name: [exact product name] Category: [what type of product it is] Key features: - [feature 1] - [feature 2] - [feature 3] [add more as needed] Target customer: [who buys this — describe in one sentence] Primary use case: [what problem does it solve, or what occasion is it for?] Selling platform: [Amazon / Shopify / own website / other] Length: [short (under 80 words) / standard (80–150 words) / long (150–300 words)] Write the full description. Lead with the benefit, not the feature. Use active voice. End with a sentence that makes the buyer feel confident about the purchase. Include the product name in the first sentence for SEO.

Write a public changelog entry for the following product update. Product: [product name] Update type: [new feature / improvement / bug fix / deprecation] What changed (technical): [describe in developer-level detail] Who is affected: [all users / specific plan / specific integration] Release date: [date] Link to docs (optional): [URL if applicable] Changelog entry format: - TITLE: [emoji] [short title — under 60 characters] - SUMMARY (2–3 sentences): what changed, why it matters to the user, and what they can do now that they couldn't before - DETAILS: one or two bullet points for the most important technical specifics - ACTION REQUIRED (if any): what the user needs to do to benefit from this change Tone: clear and factual. Avoid marketing language. Assume the reader is technical.

Format the following raw meeting notes into a clean, structured summary. Raw notes: [paste your unedited notes — bullet points, fragments, timestamps are fine] Meeting context: [meeting name, date, attendees if relevant] Audience for the summary: [who will read this — the same team / a manager / a client?] Output format: 1. MEETING PURPOSE (one sentence) 2. KEY DECISIONS MADE (bullet list — only final decisions, not discussion) 3. ACTION ITEMS (each as: WHAT → WHO → BY WHEN) 4. OPEN QUESTIONS (unresolved items that need a follow-up) 5. NEXT MEETING (if mentioned) Do not invent information not in the raw notes. If a decision or owner is unclear, mark it with [TBC] rather than guessing.

Write documentation for the following API endpoint. Endpoint: [HTTP method + path — e.g. 'POST /api/v2/orders'] Purpose: [what this endpoint does in one sentence] Authentication: [Bearer token / API key header / OAuth2 / other] Request body / parameters: [describe each field: name, type, required/optional, description, example value] Successful response: [describe the response shape: fields, types, example JSON] Error responses: [list each error code, when it occurs, and what it means] Document the endpoint in this format: - Overview (2–3 sentences) - Authentication note - Request parameters table (name | type | required | description) - Example request (curl) - Example response (JSON) - Error codes table (code | description | resolution) - Notes or rate limits if applicable

Look at this screenshot of a desktop or web application and write step-by-step instructions for completing the visible task. [attach screenshot] Target audience for the instructions: [technical user / non-technical user — pick one] Task to complete: [what the user is trying to accomplish — e.g. 'export the data as a CSV'] Level of detail: [brief (3–5 steps) / thorough (full numbered steps with UI element names)] Write the instructions as if the reader has never used this application. Use the exact names of buttons, menus, and fields as they appear in the screenshot. If a step involves clicking a specific UI element, describe it precisely (e.g. 'click the blue Export button in the top-right corner' rather than 'click Export').

Write the following code. Keep it clean, correct, and minimal — nothing extra. Language: [language + version if relevant] Task: [describe what the code must do in one or two sentences] Inputs: [what data goes in — types and examples] Outputs: [what the code returns or produces] Constraints: [no external libraries unless specified / must be async / must handle X edge case / other] When done: 1. Paste the complete, runnable code 2. Add a two-line comment above each function explaining WHAT it does and WHY (not how) 3. Identify the one input case most likely to break this code and add a guard for it

Compare these two screenshots and describe the differences. [attach Screenshot A] [attach Screenshot B] Context: [what these screenshots show — e.g. 'before and after a UI redesign', 'two versions of a landing page', 'competitor A vs competitor B'] Focus areas: [what to prioritise in the comparison — layout / typography / colour / content / spacing / navigation] Comparison output: 1. LAYOUT DIFFERENCES: how the overall structure and hierarchy changed 2. VISUAL CHANGES: colour, typography, spacing, imagery 3. CONTENT CHANGES: what text, labels, or elements appear, disappear, or move 4. UX IMPACT: which version is clearer for the user and why — one paragraph 5. RECOMMENDATION: if choosing one, which and the single strongest reason

Repurpose the following content for a different format and audience. Source content: [paste the original — article, transcript, report, post, etc.] Source format: [what the original is — blog post / podcast transcript / whitepaper / other] Target format: [what to turn it into — LinkedIn post / tweet thread / email newsletter / FAQ / press release / slide deck outline] Target audience: [who will read the output — different from the source audience? describe the shift] Tone shift (if any): [e.g. 'from technical to accessible', 'from formal to conversational'] Length target: [word count or post count] Produce the repurposed content in full. Do not add information not in the source. If the source is too thin for the target format, list what additional content would be needed to complete it properly.

Convert the data visible in this image into a clean, structured table. [attach image — chart, handwritten table, printed form, screenshot of a spreadsheet, etc.] Output format: [Markdown table / CSV / JSON array of objects / HTML table] Date or time format to use: [ISO 8601 / MM/DD/YYYY / other] Numeric format: [preserve original / convert to integers / convert to float] Missing or illegible cells: [use null / leave blank / flag with '?'] After the table, note: - Total row count extracted - Any cells flagged as uncertain and why - Any data that appeared in the image but could not be mapped to the table structure

Write a help centre article for the following topic. Product: [product name] Topic: [what the article explains — e.g. 'How to reset your password', 'Understanding billing cycles'] Target reader: [paying customer / free user / admin / developer] Prerequisites (if any): [what the reader should already know or have done before following this article] Article format: 1. TITLE: [action-oriented — starts with 'How to' or 'Understanding' or 'Setting up'] 2. OVERVIEW (2 sentences): what this article covers and who it is for 3. BEFORE YOU BEGIN (optional): any prerequisites or settings to check first 4. STEPS: numbered, each under 20 words, starting with an action verb 5. RESULT: what the reader sees or can do when they have followed all steps 6. TROUBLESHOOTING (2–3 items): the most common problems and their fixes 7. RELATED ARTICLES: [leave as a placeholder list] Tone: patient, factual, no jargon beyond what the product itself uses.

Create a slide-by-slide outline for the following presentation. Presentation topic: [what the deck is about] Audience: [who will see this — board / team / client / conference] Presentation length: [how many minutes] Goal: [what the audience should think, feel, or do after the presentation] Key information to cover: - [point 1] - [point 2] - [point 3] [add more as needed] For each slide: - SLIDE NUMBER and TITLE - VISUAL SUGGESTION (one line: what kind of visual — data chart, photo, diagram, icon grid, etc.) - SPEAKER NOTES (2–3 sentences the presenter says, not reads) - SLIDE CONTENT (bullet points or short sentences — max 6 items per slide, each under 10 words) Target: [number] slides. Last slide must be a clear call to action or next step.

Analyse these competitor screenshots and build a feature comparison table. [attach Screenshot 1 — Product A] [attach Screenshot 2 — Product B] [attach Screenshot 3 — Product C (optional)] Our product: [describe your product and its key differentiators] Comparison categories: [list the dimensions to compare — e.g. pricing, onboarding, navigation, key features visible in screenshots] Output: a Markdown table with one row per feature and one column per product After the table: 1. GAPS (features competitors have that we don't, visible in the screenshots) 2. ADVANTAGES (features we have that competitors don't, based on your description) 3. ONE RECOMMENDATION: the single highest-impact feature gap to close first, and why