Improve My Prompt

Paste a prompt you'd normally send to Claude, ChatGPT, or Gemini. Pick which research-backed enhancement layers to add — a detailed persona, step-by-step methodology, stakes framing, incentive framing, challenge framing, and confidence self-evaluation. See the before/after, copy the enhanced version.

Context and background

Read the story behind this tool: Why 'Take A Deep Breath' Makes LLMs Smarter — The Research Behind Prompt Enhancement →

Your prompt

Persona (who should the model pretend to be)

Role

Expertise (optional)

Approach (optional)

Framing layers

Step-by-step methodology

"Take a deep breath. Work through this step by step…" (OPRO — Yang et al., Google DeepMind, 2023).

Confidence self-evaluation

Model rates its own answer; refines if below threshold (Tian et al., 2023).

Stakes framing

"This is critical — getting it wrong costs…" (EmotionPrompt, Li et al., ICLR 2024).

Incentive framing ($200)

"I'll treat this as $200 of saved time" (Bsharat et al., 2023 — 26 Principles).

Challenge framing

"I bet you can't solve this cleanly — prove me wrong" (EmotionPrompt).

Strip politeness fillers

Removes "please / could you / thank you" — they don't help accuracy.

These techniques are pattern-matching tricks, not motivation — the model doesn't care about money. They do elicit higher-effort outputs in research benchmarks. Why this works — research citations →

Improve My Prompt

Context and background

Your prompt

Persona (who should the model pretend to be)

Framing layers

Before / after

Plain (baseline)

Enhanced