User Prompt Examples for Analysis, Writing, and Extraction
Summary
- Copy and adapt ready-to-use user prompt examples for analysis, writing, and extraction tasks.
- Use a simple structure (role, goal, inputs, constraints, output format) to get more consistent results.
- See prompt patterns for summarizing, comparing, critiquing, rewriting, and generating outlines and drafts.
- Get extraction prompts for tables, JSON, checklists, and validation rules (plus what to do when the model guesses).
- Learn a practical save-find-reuse workflow for prompts and snippets, including how CopyCharm can store and retrieve them.
This guide gives copy-pasteable user prompt examples you can use to analyze content, write faster, and extract structured data (tables/JSON) with fewer back-and-forth edits.
How to use these user prompt examples (so they work on the first try)
Each example below is written so you can replace bracketed placeholders like [PASTE TEXT] or [YOUR GOAL]. For best results, keep four elements explicit:
- Task: What you want (analyze, write, extract).
- Inputs: The text/data and any background context.
- Constraints: Tone, length, audience, what to avoid, what to assume.
- Output format: Bullets, table, JSON schema, headings, checklist, etc.
If you want reliability, add a short instruction like: "If a detail is missing, write 'Unknown' instead of guessing."
Prompt template you can reuse for almost anything
Use this as a base, then swap the task-specific instructions from later sections.
User prompt example (universal template):
You are [ROLE].
Goal: [WHAT YOU WANT DONE].
Context: [WHO THIS IS FOR / WHERE IT WILL BE USED].
Inputs:
[PASTE TEXT / DATA]
Constraints:
- Tone: [e.g., clear, neutral, persuasive, friendly]
- Length: [e.g., 150-200 words]
- Do not: [e.g., invent facts, mention competitors, use jargon]
- If information is missing: say "Unknown" and list what you would need.
Output format: [e.g., bullets, table, JSON, headings].
Quality check: Before finalizing, verify the output matches the requested format and constraints.
User prompt examples for analysis
Use these when you need understanding, critique, comparison, or decision support.
1) Summarize with purpose (not just a generic recap)
User prompt example:
Summarize the text below for [AUDIENCE] who needs to [DECISION/ACTION].
Include:
- 5 key points (bullets)
- 3 risks/unknowns (bullets)
- 3 recommended next steps (bullets)
If the text does not support a claim, mark it as "Unclear from the text".
Text:
[PASTE TEXT]
2) Extract arguments and evaluate strength
User prompt example:
Analyze the argument in the text below.
Output a table with columns: Claim | Evidence given | Assumptions | Counterpoints | Confidence (Low/Medium/High).
Do not add outside facts; only use what is in the text.
Text:
[PASTE TEXT]
3) Compare two options using only provided info
User prompt example:
Compare Option A and Option B using only the information I provide.
Output:
1) A side-by-side table (Criteria | Option A | Option B | Notes)
2) A recommendation for three user types: cost-sensitive, speed-focused, risk-averse
If a criterion cannot be assessed, write "Unknown".
Option A details:
[PASTE A]
Option B details:
[PASTE B]
4) Find gaps, contradictions, and missing requirements
User prompt example:
Review the text below as a requirements analyst.
Identify:
- Missing requirements (bullets)
- Contradictions or ambiguous statements (bullets)
- Questions to ask stakeholders (numbered list)
Keep it practical and specific; quote short phrases from the text when pointing to an issue.
Text:
[PASTE TEXT]
5) Turn messy notes into a clear plan
User prompt example:
Turn these notes into a structured plan.
Output sections:
- Objective
- Scope (In/Out)
- Milestones (with rough order, no dates unless provided)
- Risks and mitigations
- Open questions
Notes:
[PASTE NOTES]
User prompt examples for writing
Use these when you want drafts, rewrites, outlines, or copy variations with constraints.
1) Write an outline that is actually usable
User prompt example:
Create an outline for [CONTENT TYPE] about [TOPIC] for [AUDIENCE].
Requirements:
- Use H2/H3 headings
- Each section must include 2-4 bullet subpoints describing what to cover
- Include a short "Common mistakes" section
- Avoid filler; every section must add new information
Context/notes (optional):
[PASTE NOTES]
2) Draft from constraints (tone, length, and must-include points)
User prompt example:
Write a [LENGTH] draft for [AUDIENCE].
Topic: [TOPIC]
Tone: [TONE]
Must include these points (use all):
- [POINT 1]
- [POINT 2]
- [POINT 3]
Must avoid:
- [THING TO AVOID 1]
- [THING TO AVOID 2]
Output as: [FORMAT, e.g., email, blog section, script].
If you need assumptions, list them first under "Assumptions" and keep them minimal.
3) Rewrite for clarity without changing meaning
User prompt example:
Rewrite the text below to be clearer and easier to scan, without changing meaning.
Rules:
- Keep all factual claims intact
- Break long sentences
- Use bullets where helpful
- Preserve any numbers exactly as written
Text:
[PASTE TEXT]
4) Create multiple variations for A/B testing (without inventing facts)
User prompt example:
Create 8 variations of the copy below for the same audience and offer.
Constraints:
- Do not add new claims or features not present in the original
- Keep each variation under [X] characters/words
- Vary the angle (benefit-led, problem-led, curiosity-led, proof-led, direct)
Original copy:
[PASTE COPY]
5) Improve a draft with a targeted edit pass
User prompt example:
Act as an editor. Improve the draft below for:
- Logical flow (reorder if needed)
- Reducing repetition
- Stronger topic sentences
- Clearer calls to action (if present)
Output:
1) Revised draft
2) A short changelog (bullets) explaining major edits
Draft:
[PASTE DRAFT]
User prompt examples for extraction (tables, JSON, checklists)
Extraction prompts work best when you define the schema and tell the model what to do when data is missing.
1) Extract entities into a table
User prompt example:
Extract the following fields from the text into a table:
Columns: Name | Organization | Role/Title | Email | Phone | Location | Notes
Rules:
- Use "Unknown" if not present
- Do not guess emails/phones
- If multiple values exist, separate with semicolons
Text:
[PASTE TEXT]
2) Extract into JSON with a strict schema
User prompt example:
Convert the text into JSON that matches this schema exactly (no extra keys):
Schema:
{
"project_name": "",
"objective": "",
"stakeholders": [{"name": "", "role": ""}],
"requirements": [{"id": "", "requirement": "", "priority": "Must/Should/Could", "notes": ""}],
"risks": [{"risk": "", "impact": "", "mitigation": ""}]
}
Rules:
- If a value is missing, use "" (empty string) or [] (empty array)
- Do not invent IDs; if none exist, create IDs as R1, R2, R3 in order of appearance
- Return JSON only (no commentary)
Text:
[PASTE TEXT]
3) Extract steps and turn them into a checklist
User prompt example:
Extract all action steps from the text and convert them into a checklist.
Output format:
- [ ] Step (imperative verb first)
Group steps under headings if the text implies phases.
Do not add steps that are not explicitly supported by the text.
Text:
[PASTE TEXT]
4) Extract requirements and acceptance criteria
User prompt example:
From the text below, extract requirements and write acceptance criteria for each.
Output a table: Requirement | Acceptance criteria (Given/When/Then) | Notes/Dependencies.
Rules:
- Base acceptance criteria only on what the text implies
- If acceptance criteria cannot be derived, write "Needs clarification" and list a question
Text:
[PASTE TEXT]
5) Extract citations/quotes with location markers
User prompt example:
Extract up to 10 quotable lines from the text below.
Output a table with columns: Quote | Why it matters | Location marker.
For "Location marker", use paragraph number (P1, P2, etc.) based on the pasted text order.
Do not paraphrase; quotes must be exact.
Text:
[PASTE TEXT]
Practical prompt patterns (and when to use them)
Pattern A: "Ask for the format first"
If you keep getting messy outputs, request a proposed structure before the final answer.
User prompt example:
Before you answer, propose the best output structure for this task (headings/table/JSON) and explain why in 3 bullets.
After I confirm, produce the final output.
Task: [DESCRIBE TASK]
Input:
[PASTE INPUT]
Pattern B: "Force explicit unknowns"
When accuracy matters, instruct the model to label missing information instead of filling gaps.
User prompt example:
Use only the provided input. If a detail is missing, write "Unknown" and add it to a "Missing info" list at the end.
Task: [TASK]
Input:
[PASTE INPUT]
Pattern C: "Give a rubric, then score"
Useful for evaluating writing, resumes, proposals, or product pages.
User prompt example:
Evaluate the text using this rubric (score 1-5 for each):
- Clarity
- Specificity
- Logical flow
- Audience fit
- Actionability
Output:
1) A score table
2) The top 5 edits that would raise the score fastest (with examples)
Text:
[PASTE TEXT]
Save, find, and reuse your best prompts with CopyCharm (concrete workflow)
If you find yourself retyping the same analysis prompt, rewrite prompt, or extraction schema, saving it as a reusable snippet can help reduce repeated work. CopyCharm is a Windows desktop app that saves copied text locally, lets you search past clips, favorite important clips, and separately save reusable prompts.
A simple workflow for prompt reuse
- Save: When you write a prompt you like (for example, a JSON extraction schema), copy it and save it as a Saved Prompt in CopyCharm. When you copy a useful output (like a clean table), it can also be kept as a clip and you can Favorite it separately if it is important.
- Find: Later, search in CopyCharm using a distinctive phrase you included in the prompt (for example, "Return JSON only" or "Given/When/Then").
- Reuse: Open the saved prompt, copy it, then paste it into your AI chat, document, ticket, or email. For Claude, Gemini, Cursor, and most other apps, this is a manual copy/paste workflow.
When ChatGPT access matters: connector vs manual reuse
If you use ChatGPT and want it to pull in your saved prompt text without switching windows, CopyCharm includes an authenticated ChatGPT connector backed by optional AI Access sync. After eligible account authorization and AI Access sync, ChatGPT can search and retrieve supported Synced Data (such as Favorite Clips and Saved Prompts, plus optional Other Clips within your selected time range). ChatGPT cannot access unsynced local CopyCharm data, and retrieval is user-directed.
CTA: If you want a place to keep reusable prompts and frequently reused text you copy during work, you can try CopyCharm at https://copycharm.ai.
Common limitations (and how to prompt around them)
- Hallucinated details: If the model invents specifics, add: "Do not guess. Use 'Unknown' for missing fields."
- Inconsistent formatting: Provide a schema or an example output and require: "Match this format exactly."
- Overlong answers: Set a hard limit (words, bullets, sections) and ask it to remove anything nonessential.
- Extraction errors: Ask for a validation pass: "List any extracted fields you are not confident about and why."
- Ambiguous inputs: Ask for clarifying questions first, or request assumptions listed explicitly before the output.
Frequently Asked Questions
FAQ 1: What are "user prompt examples" and how are they different from system prompts?
Answer: User prompt examples are copy-and-edit instructions you type into a chat or tool to request analysis, writing, or extraction. System prompts are higher-priority instructions set by the platform or application that can shape behavior across a session and may not be editable in the same way by end users.
Takeaway: Use user prompts to control the task, inputs, and output format you need right now.
FAQ 2: How do I adapt these prompt examples for my exact task without making them too long?
Answer: Keep the universal structure, but only add constraints that change the outcome: audience, length, do-not-do rules, and the output format. If you are adding background, prefer a short "Context" block and paste only the relevant excerpt instead of everything you have.
Takeaway: Add constraints that matter; remove anything that does not affect the output.
FAQ 3: What should I include to get reliable analysis instead of a generic summary?
Answer: Specify the decision or purpose (what the analysis is for), require the model to separate claims from evidence, and tell it what to do with missing information (for example, label it "Unclear from the text"). Asking for a table (Claim | Evidence | Assumptions | Counterpoints) also helps keep the analysis grounded in your input.
Takeaway: Purpose + evidence boundaries + a structured format produces more useful analysis.
FAQ 4: How do I prompt for writing that matches a specific tone without adding made-up facts?
Answer: Provide the facts as bullet points (or paste the source text) and add a rule like "Do not add new claims not present in the inputs." Then define tone with a few descriptors (for example, "plain, direct, and friendly") and set a length limit. If you need creativity, constrain it to phrasing and structure, not new information.
Takeaway: Separate "facts to include" from "style to apply" and forbid new claims.
FAQ 5: What is the safest way to prompt for extraction into JSON or a table?
Answer: Define the schema explicitly, forbid extra keys/columns, and state how to represent missing values (for example, empty strings, empty arrays, or "Unknown"). Add a rule that the model must not guess, and consider asking for "JSON only" to reduce formatting noise.
Takeaway: A strict schema plus explicit missing-value rules reduces messy extraction.
FAQ 6: What do I do when the model returns the wrong format anyway?
Answer: Paste the incorrect output back and request a format-only correction: "Do not change content unless required; reformat to match this schema exactly." If it still drifts, provide a tiny example of the desired output (one row of a table or a minimal JSON object) and require it to mirror that pattern.
Takeaway: Ask for a reformat pass and provide a minimal example to anchor the structure.
FAQ 7: Can I use one prompt to analyze and then write, or should I split it into steps?
Answer: You can do either, but splitting into steps can make it easier to verify accuracy. For example: Step 1 extracts key points and constraints from your source; Step 2 writes the draft using only those extracted points. If you keep it in one prompt, require two labeled sections: "Analysis" and "Draft," and tell it not to introduce new facts in the draft.
Takeaway: Two-step workflows can be easier to check, especially when accuracy matters.
FAQ 8: How can CopyCharm help me reuse my best prompts across projects?
Answer: You can save reusable prompt text as Saved Prompts in CopyCharm, then search and copy/paste them into whatever you are working on. If you use ChatGPT, CopyCharm also offers an authenticated connector backed by optional AI Access sync; after authorization and sync, ChatGPT can search and retrieve supported synced items (and it cannot access unsynced local CopyCharm data). For other apps, reuse is manual: find the prompt in CopyCharm, then copy/paste it into the destination.
Takeaway: Save prompts once, then retrieve them by search and reuse them where you work.