Expert-level prompts in minutes, not hours.

Research, build, score and refine prompts with AI guidance, then check them against a live model.

Example run
Recommended: CoT
  1. RoleYou are a science communicator who teaches physics to young children.
  2. TaskExplain quantum superposition in under 50 words.
  3. ConstraintsKeep it simple.
  4. FormatTwo short sentences, then one question to check understanding.
0/ 100
  • Clarity-
  • Role-
  • Context-
  • Format-
  • Complexity-

Most prompts fail quietly.

They read fine, then drift, ramble or miss the point. Here is what usually goes wrong, and what changes.

  • Hours of trial and error on every promptA guided flow from research to refinement
  • No way to tell if a prompt is good enoughScores on five criteria, with the reasons
  • Losing track of what worked last weekEvery version saved, searchable and exportable
  • Guessing which reasoning technique fitsA recommendation from six reasoning frameworks

How it works

From a rough idea to a tested prompt.

Five guided steps. Each one hands the next a better draft.

Research

Enter a topic. AI researches it and recommends one of six reasoning frameworks.

Topic
Explain quantum computing to a five-year-old using metaphors
What the research found
Audience
Young children
Goal
Intuition, not precision
Shape
One idea, step by step
Chain-of-ThoughtRecommended

The explanation has one clear path. Branching would add cost without a better answer.

Also considered

Step-Back. Would help if the lesson needed the underlying principle first.

ToT. Better when several competing routes need exploring.

Ready to try it on your own topic?

Everything you need to build better prompts.

One workflow, from the first question to a prompt you can trust in production.

Research-driven insights

AI researches your topic and recommends one of six reasoning frameworks, with the reason.

Chain-of-Thought

Recommended: one clear path to the answer.

Also considered
  • ToT
  • CoT-SC
  • Step-Back
  • Grounded
  • ReAct

Guided prompt construction

Build structured prompts field by field, grounded in PDFs, spreadsheets, emails or a GitHub repo.

Role
Task
Context
Format
brief.pdf kickoff.eml your-org / api

Automated quality scoring

Five criteria, scored before and after each refinement, so you can see what improved.

94/100

AI-powered refinement

Suggested rewrites, side by side, with the reason for each change.

Keep it simple. One metaphor only: a spinning coin.

Prompt history and analytics

Every version kept, searchable and exportable, so good prompts are never lost.

  1. v158First draft
  2. v271Added role
  3. v383Format fixed
  4. v494Refined context

Six ways to make the model reason.

Different problems need different approaches. The research step recommends one for your topic, and you can always choose another.

ParsePlanSolveCheckAnswer

Chain-of-Thought

One path, reasoned step by step.

Best for

  • Structured problems with clear steps
  • Known solution paths
  • Sequential reasoning

Examples

  • Mathematical calculations
  • Code debugging
  • Financial analysis
Speed
Fast
Cost
Low

A single reasoning pass. The baseline the other ratings are measured against. Ratings are relative estimates, not benchmarks.

Not sure? The research step recommends one for your topic.

Who it’s for

Made for people who write prompts for a living.

Prompt engineers

Cut the trial-and-error loop on production prompts.

Build production-ready prompts with systematic testing and refinement.

AI researchers

Test and document reasoning approaches side by side.

Run one topic through different reasoning frameworks and compare the prompts they produce.

Developers

Ship prompts into your app with confidence.

Ground prompts in your repo, then keep every validated version on hand.

Educators

Teach prompt engineering with hands-on examples.

Guide students through a structured process with real-time feedback.

Your usage, on this device.

Analytics are stored locally in your browser and are not sent anywhere.

Numbers fill in after your first run through the workflow.

Prompts generated
0
Built with the workflow.
Prompts refined
0
Improved with the AI refiner.
Tokens processed
0
Gemini API tokens, counted locally.
Estimated API cost
$0.0000
Based on the local token count.
Unique sessions
0
Tracked in this browser.

Prompt Engineering Workflow

A guided workflow to research topics and construct sophisticated reasoning prompts.

Please Sign In

You need to sign in to access the prompt engineering workflow.

Test Run

Test your generated prompt against live models to evaluate the results.

Test Run
Beta

Generate a prompt in the workflow or load one from history to test

Prompt History

Review, inspect, and download your previously generated prompts. History is stored locally in your browser.

No Prompts Yet

Your generated prompts will appear here. Start by creating your first prompt in the workflow section above.

Build your next prompt properly.

Bring your Gemini API key. Your prompt history stays in this browser.