Guides

Prompt Lab

How your prompts are scored and improved.

  • Measure a skill every 2026 employer expects: reliably great LLM output
  • See a dimension-by-dimension score, not just a pass/fail
  • Learn from an optimized rewrite of your own prompt, with rationale

Learn to drive an LLM

You're given a task; you write a prompt; Miatz runs it and then scores the prompt against a rubric (clarity, context, constraints, output format, reasoning, grounding). You get a dimension breakdown, the techniques you used, and an optimized rewrite with rationale.

Why it matters

In 2026, "can you get great output from an LLM, reliably?" is a core engineering skill. The Prompt Lab measures and grows exactly that.

The journey, step by step

StepWhat you doWhat you get
Get a taskRead the task the lab gives youA concrete goal your prompt must achieve
Write a promptDraft the prompt you'd send an LLMMiatz runs it for real
Get scoredReview the rubric breakdownScores for clarity, context, constraints, output format, reasoning and grounding
Study the rewriteCompare your prompt to the optimized rewriteYou see exactly which techniques lift the score, and why

Frequently asked

How is my prompt scored?

Against a rubric of clarity, context, constraints, output format, reasoning and grounding — with a dimension breakdown, the techniques you used, and an optimized rewrite with rationale.

Does the lab actually run my prompt?

Yes. Miatz runs your prompt and scores both the prompt and the outcome, so the feedback reflects real LLM behavior.

Why does prompt engineering matter?

Getting great output from an LLM reliably is a core engineering skill now. The Prompt Lab measures and grows exactly that.

Related guides

Was this helpful?

Try it, don't just read about it

Founding cohorts are free. Take the DSAT and see this from the inside.