šØ Anthropic dropped a free guide on Prompt Engineering and it's insanely useful.
Learn how to write smarter prompts that get better results.
Hereās whatās inside: https://t.co/s2KuiGYDhF
1. Prompt generator
Learn how to get Claude to write prompts for you.
Itās meta and wildly effective.
2. Use prompt templates
Save time.
Write once, reuse forever.
Templates = scalable prompting.
3. Prompt improver
Claude can improve your own prompts.
Turn āmehā into āmagic.ā
Itās like Grammarly for AI instructions.
4. Be clear and direct
Vague prompts = vague results.
Clear prompts = magic.
The #1 rule of prompting? Say exactly what you want.
5. Use examples (multishot prompting)
Show Claude how to think.
Multiple examples = tighter, smarter responses.
Think: "Learn by showing."
6. Let Claude think (CoT)
Donāt rush the answer.
Ask Claude to think step-by-step.
Chain-of-thought = next-level reasoning.
7. Use XML tags
Control Claudeās outputs with precision.
Think structure.
Think clarity.
Think clean results.
8. Give Claude a role (system prompts)
"You're a lawyer."
"You're a therapist."
Assigning roles = instant perspective shift.
It works.
9. Prefill Claude's response
Want a certain tone or structure?
Start the reply for Claude.
Guides output exactly how you want.
10. Chain complex prompts
One prompt isn't always enough.
Break tasks into steps.
Each output feeds the next.
11. Long context tips
Claude can handle big docs.
But you have to feed it right.
These tips show how to do that without overload.
12. Extended thinking tips
Make Claude think harder.
Ask better follow-ups.
Itās not just what you ask itās how you iterate.
I hope you've found this thread helpful.
Follow me @chatgptricks for more.
Like/Repost the quote below if you can: