AI Prompt Library
Find practical prompt templates for marketing, support, research, coding, and structured AI workflows. Copy a prompt, fill in the variables, and adapt it for ChatGPT, Claude, Gemini, or your own AI stack in 2026.
Build Better AI Outputs From Reusable Patterns
The library focuses on clear roles, explicit context, structured outputs, verification steps, and reusable variables so prompts stay useful across models and teams.
What separates a saved prompt from a throwaway prompt?
Strong prompts are not longer by default. They make the desired output, source context, constraints, and review criteria explicit enough that the same workflow can be repeated by another person or another model.
Role + goal Context + source material Constraints + audience Output format Review checklist
Outcome first
State the exact deliverable before adding background context: brief, table, plan, code diff, QA checklist, or decision memo.
Context with boundaries
Include audience, channel, constraints, source material, and what the model should ignore or avoid.
Structured output
Ask for a format that can be reviewed quickly: headings, bullets, JSON, markdown tables, acceptance criteria, or ranked options.
Verification pass
Add a final review step for assumptions, missing data, risk flags, citations, edge cases, or next questions.
Start from the job, not the model
The fastest path is usually to pick the work pattern first, then tune the prompt for the model, tone, and source material you already have.
Marketing launch
Campaign ideas, landing-page copy, emails, positioning, personas, and social posts.
Support operations
Refund macros, escalation replies, customer summaries, policy-safe responses, and FAQ drafts.
Developer work
Code review, test planning, debugging, documentation, refactoring, and implementation checklists.
Research synthesis
Source comparison, executive summaries, competitor analysis, decision matrices, and evidence reviews.
Adapt the same prompt across AI tools
Most useful prompts are portable, but the best version changes with model strengths, context length, and whether the task is creative, factual, or operational.
Works well with explicit roles, examples, and iterative refinement. Ask for a first draft, then request critique and revision.
Strong for long-context review, editorial judgment, and nuanced rewrites. Put source material and decision criteria up front.
Useful for multimodal and research-heavy workflows. Ask it to separate observed facts from inferred recommendations.
Keep instructions shorter, reduce nested tasks, and request smaller intermediate outputs before a final synthesis.
Prompt rules worth bookmarking
These patterns are distilled from current OpenAI, Anthropic, Google, and Microsoft guidance, then translated into practical checks for everyday work.
Decide what a good answer looks like and test prompts against real examples before polishing wording.
Tell the model what to do when the request is ambiguous, out of scope, missing context, or incompatible with the requested format.
Few-shot examples are useful when the output needs a consistent tone, schema, label set, or decision pattern.
For tables, JSON, or checklists, keep the format small, named, and easy to validate instead of asking for everything at once.
Ask for extraction, analysis, critique, and final synthesis separately when one prompt becomes too broad.
Different model families and versions can respond differently, so durable prompts need periodic review.
Keep these three prompts in your back pocket
Use these as starting points when no library card fits exactly. Replace the bracketed fields, paste your source material, and keep the review step intact.
Universal task prompt
Good default for writing, planning, analysis, and operations work.
Act as [ROLE]. Goal: [WHAT SHOULD BE PRODUCED] Audience: [WHO WILL USE THIS] Context: [BACKGROUND, SOURCE MATERIAL, CONSTRAINTS] Requirements: - Use [TONE / FORMAT / LENGTH]. - Prioritize [SUCCESS CRITERIA]. - Do not [BOUNDARIES OR EXCLUSIONS]. Before the final answer, check for missing assumptions, unclear inputs, and risks. If critical information is missing, ask up to 3 questions first.
Research synthesis prompt
Useful when summarizing sources, competitors, market notes, or technical docs.
Review the source material below and create a grounded synthesis. Question to answer: [QUESTION] Source material: [PASTE NOTES, LINKS, TRANSCRIPTS, OR EXCERPTS] Return: 1. Direct answer in 3-5 bullets 2. Evidence table with claim, source, confidence, and caveat 3. Contradictions or missing information 4. Practical next steps Separate facts from interpretation. If the source material does not support a claim, say so.
Quality review prompt
Paste after a draft to improve accuracy, usefulness, and fit.
Review this draft for [AUDIENCE] and [GOAL]. Draft: [PASTE DRAFT] Score from 1-5: - Clarity - Specificity - Evidence - Actionability - Risk or compliance issues Then provide: 1. The 5 most important fixes 2. A revised version 3. What you changed and why 4. Remaining assumptions to verify
Official prompting references
For deeper prompt tuning, start with the model provider docs rather than viral prompt packs.
Model-specific prompting, roles, evaluations, and iteration guidance.
Success criteria, evals, clarity, examples, role prompting, and prompt chaining.
Clear instructions, iterative prompt design, and Gemini-oriented strategies.
Prompt components, few-shot learning, grounding, and validation reminders.
Structured output behavior, incompatible inputs, mistakes, and schema hygiene.