ReframeStep 1
Layer your prompt
Each lesson unlocks a new section — identity, principles, strategies, examples. You build up, one layer at a time.
How it works
Build it. Run it. Measure it.
Reframe teaches prompt engineering through a feedback loop you can see. Every lesson, every layer, every score delta.
The Loop
Three steps. Every lesson.
Layer, run, compare. Repeat.
Step 2
Run it live
Your prompt drives an AI agent through Word Guess in real time. Every decision traces back to what you wrote.
Step 3
Measure the difference
Compare before and after, side by side. Solve rate, average guesses, total score — you see exactly what your change was worth.
Module 1: Prompt Anatomy
Five layers of a great prompt
Each lesson unlocks one. By the end, you've built the whole thing.
Lesson 1
Task
The objective. Always present, always locked. Tells the agent what it's playing.
Lesson 2
Identity
A role that anchors reasoning. One sentence changes how the model thinks.
Lesson 3
Principles
Guardrails for judgment. Rules that prevent lazy, random decisions.
Lesson 4
Strategies
A playbook of specific moves. Tactics that translate principles into action.
Lesson 5
Examples
Worked demonstrations. Show the model what good looks like.
Why Reframe
Not another tutorial.
Real runs. Real scores. Real feedback.
Every change, measured
Each lesson runs your prompt and shows the score delta. You always know what your edit was worth.
Side-by-side comparison
Your old prompt vs. your new one, running in parallel. The numbers speak for themselves.
Free play when you're ready
Take what you learned to the playground. Write any prompt, enter the weekly challenge, climb the leaderboard.