Vibecoding
Build for the vibe, debug later
Start new thread
Ilai Szpiezak

Prompt Engineering or Context Engineering?

by
Pretty Prompt
•

Hey PH community!

I’ve been spending a lot of time thinking about how people actually work with prompts while building a tool in this space, and I realized I have way more questions than answers.

A few I’d love to throw out:

  1. Where do your “good” prompts live? Do you keep a library somewhere, or do you mostly keep them inside old chats?

  2. Do you test your prompt with different models?

  3. How do you know if you've missed anything in your prompt? Ask ChatGPT, or iterate somewhere else?

  4. And how do you think about Context when vibe coding?

Please share your experience below! This is super helpful! 🙌

5 views

Add a comment

Replies

Best
Alper Tayfur

Great questions — this is exactly where the conversation is shifting.

For me, it’s way more context engineering than prompt engineering now. Prompts matter, but they’re brittle. Context is what makes things reliable.

  • My “good” prompts usually end up living outside chats — in docs or repos — but what really matters is the inputs around them (examples, constraints, schemas).

  • I rarely test the same prompt across many models unless it’s mission-critical. Most breakages come from missing context, not model choice.

  • If something feels off, I don’t ask the model “what’s wrong with the prompt.” I add structure: clearer goals, examples, or step-by-step framing.

  • When vibe coding, context is everything: existing code, naming conventions, constraints, and what not to touch. Without that, even a great prompt drifts fast.

Big shift for me has been this:
prompts are instructions, context is the environment.
If the environment is right, the prompt can be pretty simple and still work well.

Peter Claridge
StreamAlive - Interactive PPT slides

I'm a daily user of chatgpt, claude and recently Google Antigravity. Both for work and on personal projects.

  1. Where do your good prompts live? I have no idea whether my prompt is a good prompt or not. I seem to get the desired output so haven't found the need to save or remember anything.

  2. Testing prompts with new models? No...

  3. Missing anything in the prompt? I just edit the initial prompt if I realize I forgot something in the prompt.

  4. Context in vibe coding - not sure what that means?

I imagine I am chaos when it comes to prompting 😂 but haven't found a need to structure what I ask an LLM, they all seem to understand what I'm asking, but also I can't remember I time when I've had to repeat a prompt.