Smart Image Optimizer
Compress images with adjustable quality and resize options. Batch processing, side-by-side comparison, and local privacy — no uploads needed.
Open ToolKeyboard Shortcuts
Prompt Engineering Tool
Build, analyze, enhance, and save AI prompts. Craft perfect prompts for ChatGPT, Claude, Gemini, and any LLM.
The output you get from an AI depends on the prompt you give it. Vague in means mediocre out. This tool helps you build prompts with structure — role, context, task, format, and constraints — so you get useful results instead of generic fluff. Templates cover common scenarios, and the analyzer scores your prompt and points out what's missing. Everything runs locally, nothing leaves your browser.
1. Choose a Role
Define the AI's persona and expertise level.
2. Define the Task
What should the AI do?
0 words
3. Add Context
Provide background info and specific requirements.
The more context, the better the AI's response.
4. Set Output Format
Control how the AI structures its response.
5. Advanced Options
Fine-tune how the AI approaches your task.
Generated Prompt
Analyze a Prompt
Paste any prompt below to get a quality score and improvement suggestions.
0 words / 0 characters
Quality Scores
Suggestions
Enhance a Basic Prompt
Enter a simple prompt and we will automatically improve it with best practices: clear role, structured output, examples, and constraints.
Enhanced Prompt
A/B Comparison
Prompt Templates
Click a template to load it into the builder. You can also save your own prompts as templates from the Enhance tab.
Recently Used
My Saved Prompts
No saved prompts yet
Build or enhance a prompt and click Save to add it here.
Quick Answer
Prompt Engineering Tool is a Build, analyze, and refine AI prompts with quality scoring, structured templates, and actionable improvement suggestions. by ThinkaBell. It runs all processing happens locally in your browser — no sign-up, no file uploads, and no data ever leaves your device.
- Feature: Build, analyze, and refine AI prompts with quality scoring
- Feature: Structured templates with actionable improvements
- Feature: Privacy-first — all processing runs in your browser
- Feature: Free to use with no account or registration required
- Privacy-first: no data collection, no tracking, no uploads.
- Free forever: no premium tiers, no hidden charges, no accounts.
Key Terms
- Web Application
- A web application that runs entirely in your browser using HTML, CSS, and JavaScript. No software installation, no account creation, and no file uploads to external servers.
- Client-Side Processing
- All calculations and data processing happen on your device. Your information never leaves your browser, ensuring complete privacy and security.
- AI Tools
- A utility accessible through a web browser at no cost. ThinkaBell tools are free with no premium tiers, no usage limits, and no account requirements.
Related Tools You Might Find Useful
These tools complement what you're using and solve similar tasks.
Study Assistant
Turn any text into summaries, key points, flashcards, quizzes, or study plans — five study formats from one input.
Open Tool
Blog Post Generator
Generate a dynamic blog outline from any topic with varied angles, SEO analysis, and export options.
Open Tool
CV Generator
Build a professional resume from three templates (Classic, Modern, Minimal) with live preview, ATS scoring, and PDF/DOCX/TXT export.
Open ToolStay updated on new tools
Get a short email when we launch new free tools. No spam, unsubscribe anytime.
Prompt Engineering Techniques: A Complete Guide
Master these proven techniques to get better results from any AI language model. Each method has a specific use case and works with ChatGPT, Claude, Gemini, and similar models.
Zero-Shot Prompting
The simplest technique: ask the AI to perform a task without providing examples. Best for straightforward tasks the model already understands well. Example: "Write a haiku about machine learning." Works because modern LLMs have broad training coverage.
Few-Shot Prompting
Provide 2-5 examples of desired input-output pairs before asking the real question. This dramatically improves accuracy for formatting, style, and reasoning tasks. Example: Show three examples of customer emails with proper responses, then ask for the fourth.
Chain-of-Thought (CoT)
Ask the AI to think step by step before giving the final answer. This reduces errors on complex reasoning tasks like math problems, logic puzzles, and multi-step analysis. Simply add "Let's think step by step" or "Walk through your reasoning" to your prompt.
Role Prompting
Assign a specific persona to the AI before the task. A role definition sets expertise level, vocabulary, and response style. Example: "You are a senior cardiologist explaining heart disease to a patient." The Build tab in this tool lets you select roles easily.
Structured Output Prompting
Specify exactly how the AI should format its output: markdown, JSON, CSV, tables, or bullet points. This makes responses machine-readable and easier to parse. Always specify format constraints in your prompt.
Constraint-Based Prompting
Set explicit boundaries: word count limits, topics to avoid, reading level, or formatting rules. Constraints prevent the AI from going off-topic and keep responses within usable parameters.
Quick Reference: Prompt Formulas
Copy these proven formula templates and fill in your details.
APSE Framework
Act, Persona, Specifics, Examples
Act as [PERSONA]. Your task is to [TASK]. Be specific about [DETAILS]. Here is an example: [EXAMPLE]
RTF Format
Role, Task, Format
Role: [PERSONA] Task: [WHAT TO DO] Format: [OUTPUT STRUCTURE]
CARE Model
Context, Action, Result, Example
Context: [BACKGROUND] Action: [WHAT TO DO] Result: [DESIRED OUTPUT] Example: [SAMPLE]
Chain-of-Thought
Step-by-step reasoning
[QUESTION] Walk through your reasoning step by step. Explain your thinking at each stage. Then provide the final answer.
Prompt Planning Worksheet
Fill out this worksheet before building your prompt. It helps you organize your thoughts and ensures nothing is missed. Your answers are saved in your browser.
Common Prompt Mistakes and How to Fix Them
Even experienced users make these mistakes. Recognizing them is the first step to writing better prompts.
Being Too Vague
Mistake: "Write about climate change." This gives the AI no direction on angle, audience, depth, or format. Fix: "Write a 500-word explainer on climate change for high school students. Focus on three main causes and their effects. Use simple language and include one real-world example for each cause." The difference is specificity.
No Role or Persona
Mistake: Asking for advice without establishing who the AI should be. This produces generic, surface-level responses. Fix: "You are a seasoned startup founder who has raised two rounds of venture capital. Give me advice on preparing a pitch deck for seed funding." The role primes the AI to draw from relevant expertise.
Missing Output Constraints
Mistake: Not specifying format, length, or tone. The AI might write a 2000-word essay when you needed three bullet points. Fix: Always include constraints: "Respond in three bullet points. Keep each bullet under 50 words. Use a professional tone." Use the Format and Tone dropdowns in the Build tab to set these automatically.
Overloading the Prompt
Mistake: Asking the AI to do too many things in one prompt: "Explain quantum computing, write a poem about it, and create a study guide." The AI will spread its attention thin. Fix: Split complex requests into separate prompts. Use the My Prompts feature to save each one. One task per prompt yields significantly better results.
No Context or Background
Mistake: "Explain how REST APIs work" without mentioning your current knowledge level. The AI might use jargon you do not understand or explain basics you already know. Fix: "I am a junior frontend developer who knows JavaScript but has never built an API. Explain REST APIs, assuming I understand HTTP methods but not server architecture." The AI adjusts accordingly.
Not Iterating
Mistake: Accepting the first response as final. The first output is rarely the best. Fix: Use the Analyze tab to check your prompt's quality. Then enhance it with more context, better constraints, or a clearer role. Great prompts are built through iteration, not perfected in one try.
Prompt Use Cases by Profession
Different roles benefit from different prompting approaches. Here are tailored strategies for common professions.
For Students and Researchers
Use the AI as a research assistant and tutor. Ask it to summarize papers, explain difficult concepts, generate study questions, and help structure essays. The Teacher role works well here. Always verify facts against primary sources since AI models can make errors. Combine with the ThinkaBell Citation Generator to properly cite any AI-assisted research.
For Software Developers
Use the Developer role and be specific about your tech stack. Include framework versions, coding standards, and error messages. For code generation, request inline comments and error handling. For debugging, paste the error and describe what you have tried. The Code Reviewer template is particularly useful for catching security issues and performance problems.
For Marketers and Content Creators
Use the Marketing Professional role for campaign ideas, ad copy, and content strategy. Specify your brand voice, target audience, and platform. The Blog Post Writer and Social Media Content templates are excellent starting points. For SEO, ask for keyword-optimized headings and meta descriptions. Always customize AI-generated content to match your brand's unique voice.
For Writers and Editors
Use the Writer role for drafting, the Researcher role for fact-checking, and the Editor role for polishing. The Analyze tab can evaluate your own writing for clarity and structure. Use the Rewrite task type to improve existing text. For creative writing, the Story Generator template provides a solid framework. Treat AI output as a first draft that you refine.
For Business Professionals
Use the Strategist role for business analysis, the Analyst role for data interpretation, and the Consultant role for recommendations. The Business Plan Outline and Competitor Analysis templates structure complex business thinking. For meetings, paste transcripts into the Meeting Notes Summarizer template. Always review AI outputs for accuracy before sharing with stakeholders.
Model-Specific Prompting Tips
Different AI models have different strengths. Adapt your prompts to get the best results from each one.
ChatGPT
Excels at creative writing, brainstorming, and conversational tasks. Strong at following multi-step instructions. Best with clear, structured prompts. Use the Conversational tone for dialogue tasks. For code, be explicit about language version and framework. ChatGPT handles role-playing particularly well, making it ideal for the Role prompting technique.
Claude
Known for careful reasoning, long-form analysis, and safety-conscious responses. Performs exceptionally well with Chain-of-Thought prompting. Claude excels at document analysis, detailed writing tasks, and nuanced discussions. It is particularly strong at following format instructions and producing well-structured long-form content.
Gemini
Strong at factual accuracy and connecting information across domains. Works well with structured output formats like tables and JSON. Gemini handles multimodal inputs effectively and is particularly good at tasks requiring precise, well-organized responses. Use the Technical tone for maximum precision in factual responses.