All templates
⚡ Advanced Techniques·Works on: claude, chatgpt, gemini, grok
Chain-of-Thought Forced Reasoning
Technique: Chain-of-Thought (CoT)
Make the model show its work before answering — dramatic accuracy gain on multi-step problems.
Advanced#reasoning#accuracy#math#logic
0/1
Fill the template
1 placeholder left.
~231tokens
Live preview· click any pill to jump to its field
Act as a careful problem-solver. I will give you a complex question. Do NOT answer immediately. Problem: Follow this exact protocol: 1. **Restate the problem** in your own words in one sentence. (Verifies you understood it.) 2. **List what you know** — every fact in the problem, as bullet points. 3. **List what you don't know** — every unknown, as bullet points. 4. **Identify the bridge** — the relationship that connects knowns to unknowns. 5. **Reason step-by-step** — number each step. At each step, state what you're computing/deducing and why this is the right next step. 6. **Sanity check** — pick ONE step and verify it with a different method. 7. **State the answer** in one sentence. 8. **Confidence**: low / medium / high, with the strongest reason you might be wrong. If at step 4 you realise the problem is under-specified, STOP and ask the clarifying question instead of guessing.
Inputs0 of 1
›See the lazy version this template replaces
Before — the lazy prompt
What's the answer to <complex question>?
Why it works
- Forcing explicit steps catches arithmetic and logic errors that intuition would miss.
- The 'what you don't know' bullet is the most-skipped step in human reasoning — and where most errors originate.
- The sanity check with a different method is the single move that doubles answer accuracy on complex problems.
- Permission to stop and ask kills the 'confidently wrong' failure mode.
Make this one yours
Replace the bracketed placeholders, then paste into the Prompt Fixer to lint your customisation before hitting send.