How to Start a Prompt Library With Your First 20 Prompts

Summary

• Start your prompt library with 20 prompts that cover repeatable work: clarify, plan, draft, revise, evaluate, and hand off.
• Store prompts as reusable building blocks (not one-off chats) by separating: role, inputs, constraints, and output format.
• Use a simple naming convention and a short “when to use” line so you can find the right prompt quickly.
• Improve prompts by logging what changed (inputs, constraints, examples) and saving the better version as a new prompt.
• CopyCharm can help you save prompts and retrieve them later; with its authenticated ChatGPT connector, ChatGPT can search/retrieve only supported synced items after authorization and sync.

A prompt library is just a personal (or team) collection of prompts you can reuse. The fastest way to start is not to chase “perfect prompts,” but to capture the first 20 that cover your most repeated tasks. Once you have those, you can iterate: add inputs, tighten constraints, and standardize outputs.

This guide gives you a practical structure, a set of 20 starter prompts you can copy, and a workflow for saving, finding, and reusing them across ChatGPT, Claude, Gemini, Cursor, and your day-to-day tools.

What counts as a “prompt library” (and what doesn’t)

A useful prompt library has prompts that are:

• Reusable: they work across multiple projects with only input changes.
• Findable: you can search by task (“rewrite,” “summarize,” “screen candidates”).
• Consistent: they ask for outputs in a predictable format (bullets, table, JSON, email draft).
• Upgradable: you can refine them without losing the older version that still works.

What doesn’t count: a long chat thread you can’t reliably reuse, or a prompt that only makes sense with hidden context you forgot to save.

The simplest structure for every prompt (copy this template)

Use this lightweight template for almost every prompt you save. It keeps prompts short, but makes them easier to reuse.

Your first 20 prompts (a starter library you can actually use)

These 20 are designed to cover common knowledge-work loops: clarify the task, plan the work, produce a draft, improve it, and package it for someone else. Replace bracketed text with your inputs.

Category A: Clarify and scope (Prompts 1-4)

• 1) “Ask me the missing questions”
  Prompt: “Before you answer, ask up to 10 clarifying questions. Prioritize questions that change the approach. Here’s what I know so far: [context]. My goal is: [goal]. Constraints: [time/budget/tools].”
• 2) “Turn a messy request into a spec”
  Prompt: “Convert this request into a clear spec with: objective, audience, success criteria, constraints, assumptions, and open questions. Request: [paste].”
• 3) “Define done”
  Prompt: “Create a ‘definition of done’ checklist for this task. Include quality checks and edge cases. Task: [paste].”
• 4) “Risk and dependency scan”
  Prompt: “List risks, dependencies, and failure modes for this plan. For each, propose a mitigation. Plan/context: [paste].”

Category B: Research and synthesis (Prompts 5-8)

• 5) “Summarize for a specific audience”
  Prompt: “Summarize the text below for [audience]. Keep it to [length]. Include: key points, implications, and recommended next steps. Text: [paste].”
• 6) “Extract entities and facts”
  Prompt: “Extract a table with columns: Entity, Claim, Evidence in text (quote), Confidence (low/med/high). Text: [paste].”
• 7) “Compare options”
  Prompt: “Compare these options using criteria: [criteria]. Output a table plus a recommendation for [use case]. Options: [paste list]. Context: [paste].”
• 8) “Turn notes into an outline”
  Prompt: “Turn these notes into a structured outline with headings and bullet points. Keep gaps explicit as ‘TBD’. Notes: [paste].”

Category C: Writing and rewriting (Prompts 9-12)

• 9) “Draft from outline”
  Prompt: “Write a first draft from this outline. Audience: [audience]. Tone: [tone]. Constraints: [length/claims]. Outline: [paste].”
• 10) “Rewrite for clarity”
  Prompt: “Rewrite the text to be clearer and more direct. Keep meaning. Remove fluff. Provide: (1) revised version, (2) 5 key edits explained. Text: [paste].”
• 11) “Shorten without losing meaning”
  Prompt: “Reduce this to [target length]. Preserve key details and any numbers. Output: final version + a bullet list of what you removed. Text: [paste].”
• 12) “Tone shift”
  Prompt: “Rewrite for tone: [friendly/professional/firm/empathetic]. Keep the same facts. Provide 2 variants. Text: [paste].”

Category D: Quality control (Prompts 13-16)

• 13) “Critique against criteria”
  Prompt: “Evaluate this output against these criteria: [criteria]. Give a score (1-10) per criterion and specific fixes. Text: [paste].”
• 14) “Find contradictions and missing steps”
  Prompt: “Identify contradictions, missing steps, and ambiguous statements. Suggest exact rewrites. Text: [paste].”
• 15) “Compliance and sensitivity check (non-legal)”
  Prompt: “Flag statements that could be risky, misleading, or too absolute. Suggest safer wording. Context: [industry/audience]. Text: [paste].”
• 16) “Create test cases / edge cases”
  Prompt: “Generate edge cases and test cases for this logic/process. Output as a table: Case, Input, Expected outcome, Notes. Process: [paste].”

Category E: Packaging and handoff (Prompts 17-20)

• 17) “Executive summary + next steps”
  Prompt: “Write an executive summary (max [X] words) and a next-steps list with owners and timelines (placeholders ok). Content: [paste].”
• 18) “Meeting agenda + pre-read”
  Prompt: “Create a 30-minute meeting agenda for this goal. Include pre-read, decisions needed, and questions to resolve. Goal/context: [paste].”
• 19) “Email draft”
  Prompt: “Draft an email to [recipient] about [topic]. Tone: [tone]. Include: context, request, deadline, and a short TL;DR. Details: [paste].”
• 20) “Create a reusable SOP”
  Prompt: “Turn this process into an SOP with steps, inputs, outputs, checks, and common failure points. Process/notes: [paste].”

How to tailor the first 20 prompts to your role (fast swaps)

You do not need separate libraries for every job title. Keep the same 20 “shapes,” and swap the evaluation criteria and output formats.

• Consultants: add “success criteria,” “assumptions,” and “client-ready formatting” to prompts 2, 13, 17.
• Marketers: add “audience segment,” “offer,” “objections,” and “CTA variants” to prompts 9-12.
• Recruiters: add “must-have vs nice-to-have,” “screening questions,” and “red flags” to prompts 6, 7, 16.
• Researchers/analysts: add “definitions,” “limitations,” and “what would change my mind” to prompts 5-8, 13.
• Developers: add “constraints,” “interfaces,” “acceptance criteria,” and “test cases” to prompts 2, 16, 20.
• Content teams: add “style rules,” “voice,” and “fact boundaries” to prompts 9-15.
• Support teams: add “empathy,” “policy boundaries,” and “next action” to prompts 12, 19.
• Ecommerce operators: add “SKU attributes,” “benefits vs features,” and “returns/shipping constraints” to prompts 9-12, 15.

Naming and organizing: a lightweight system you can keep up with

Your library will only get used if you can find things quickly. A simple convention is enough:

• Name format: “Verb + deliverable + audience” (e.g., “Summarize call notes for client email”).
• First line: “When to use: …”
• Include the output format: “Output as a table with columns …”
• Keep variants explicit: “Rewrite - firm tone (v2)” rather than overwriting the original.

If you work across multiple AI tools (ChatGPT, Claude, Gemini, Cursor), keep prompts tool-agnostic where possible. When a prompt depends on a tool feature (like a specific file upload workflow or a project workspace), save a note in the prompt itself: “Tool note: requires [feature].”

A concrete workflow in CopyCharm: save, find, and reuse prompts (and clips)

If you are building a prompt library while doing real work, you will repeatedly copy drafts, snippets, and prompts between tools. CopyCharm is a Windows desktop app that saves copied text locally, lets you search past clips, favorite important clips, and separately save reusable prompts.

Workflow: capture your first 20 prompts while you work

• Save: When you write a prompt that worked (for example, “Turn a messy request into a spec”), save it as a Saved Prompt in CopyCharm. When you copy a useful output (like a polished email or a checklist), it can be stored as a clip; you can Favorite the important ones so they are easier to return to later.
• Find: Later, search in CopyCharm using the verb you remember (“rewrite,” “summarize,” “SOP,” “screen”). This is why the naming convention matters.
• Reuse: Copy the saved prompt (or a favorited clip) and paste it into the tool you are using (Claude, Gemini, Cursor, email, docs). This is a manual cross-tool workflow: search/retrieve in CopyCharm, then copy/paste into the destination app.

When ChatGPT access matters: authenticated connector + synced-data boundary

If you want ChatGPT to help you locate a prompt you saved, CopyCharm also has an authenticated ChatGPT connector backed by optional AI Access sync and a read-only MCP service. After you sign in with the account for an eligible active CopyCharm purchase, authorize the CopyCharm Desktop connection, enable and complete AI Access sync, and authorize the ChatGPT connector, ChatGPT can search or list supported synced items and retrieve a selected item’s full text.

Important boundary: ChatGPT can search and retrieve only supported Synced Data (in the categories you enabled for sync). It cannot access unsynced local CopyCharm data.

What to sync (and what not to)

AI Access sync is optional and scoped. You can enable supported categories such as Favorite Clips and Saved Prompts, and optionally Other Clips within a selected time range. Other Clips are off by default, and general clipboard history is not automatically uploaded. CopyCharm also filters detected Restricted Data before AI Access sync; screenshot images, image bytes, and local file paths remain local.

This setup can be useful if your goal is: “I want ChatGPT to help me find the prompt I saved last week,” while still keeping your broader clipboard history as a local desktop workflow.

Try CopyCharm for building your prompt library on Windows: https://copycharm.ai

How to improve prompts without over-engineering them

After you use a prompt 3-5 times, upgrade it with one change at a time:

• Add a better output format: “Output as a table with columns …”
• Add constraints: length, tone, “do not invent facts,” “ask questions if missing inputs.”
• Add an example: one short example input and the desired shape of output.
• Add a “stop condition”: “If you cannot answer without assumptions, list assumptions and ask me to confirm.”

Save the improved version as a new prompt (v2) so you can fall back if the new constraints are too tight for a different situation.

Common mistakes when starting a prompt library (and quick fixes)

• Mistake: Saving prompts with missing inputs.
  Fix: Add an “Inputs:” block that tells future-you what to paste.
• Mistake: Prompts that are too long to reuse.
  Fix: Move long context into a separate reusable “context block” you can paste when needed.
• Mistake: No consistent output format.
  Fix: Decide on 2-3 formats you like (bullets, table, JSON) and standardize.
• Mistake: You can’t find anything later.
  Fix: Rename prompts with verbs and outcomes; keep “when to use” as the first line.
• Mistake: Treating one tool’s chat history as your library.
  Fix: Save prompts outside the chat so you can reuse them across tools and projects.

Frequently Asked Questions

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Related Guides

• What Is an AI Context Pack?
• How to Prepare Better Context for ChatGPT
• How to Organize Copied Text for AI Work
• Source-Labeled Context for AI: Why It Matters
• ChatGPT Context Pack for Consultants