Reddit trick: ask the AI to write the one-shot prompt after a chat finally works
Ok_Negotiation_2587 · reddit · 2026-10-13
A Reddit user shared a prompt engineering trick: when a conversation finally gets the right result after ten messages of back-and-forth, ask the model to reverse-engineer the one-shot prompt.
The instruction: "This conversation finally got to what I wanted. Write the single prompt I should have sent at the start to get this result in one reply. Include the context, constraints and format we only figured out along the way. Put anything that will change next time in {{doublebraces}}. Then list the questions you'd still need to ask me if I used this prompt cold."
The final step matters most — the model's questions reveal what context is still missing. Saved prompts with {{ }} blanks become reusable templates; even a notes file with five of these beats re-running the ten-message dance. The author also plugs his own extension for managing such prompts across ChatGPT, Claude, Gemini and Grok.
More from Apps
- 3 devs spent 2 years building an AI-assisted roguelike horror card game, demo on Steam — AlvinhoGames_ · 2026-10-23
- 3 flight-booking rules my AI agent checks that I'd never do by hand — Efistoffeles · 2026-10-23
- OmarchyTV beta brings DHH's Omarchy-style launcher to Android TV — alexcovo_eth · 2026-10-23
- Perceptic launches PercepticOS, an AI workspace for 'scientific superintelligence' — nathanbenaich · 2026-10-23
- Capy 0.5.0 lets Claude Pro/Max subscribers reuse their plan quota, adds OIDC cloud access — garrytan · 2026-10-23
- ChatGPT + HeyGen MCP turn job listings into video cover letters in your face and voice — HeyGen · 2026-10-23