#User Interfaces

1 messages · Page 1 of 1 (latest)

gloomy crowBOT
#

UI Technology Thread

agile osprey
#

One avenue I'm exploring to keep work sessions structured is the use of remote prompting.
Note: The ‘remote prompt’ is just a trick anyone can make—no special tools required. Have fun with it...

Instead of embedding all the business logic directly into the interface, I reference an external definition file (such as TiSpI/NtreeS) and ask the AI to execute a specific index.
This seems to promote a better separation between the UI "grammar" and the conversation flow. Is this an approach you use for your component testing, or do you prefer direct context injection?

Is this method already well-known?

How to use remote prompting?

  1. Open Gemini AI.
  2. In the prompt: enter the link to the remote file.
  3. Then type: "Please respond to prompt #N".
    Advantage: You don't even need to open the remote page yourself.

Example:

Your prompt in Gemini:
https://www.wvanim.fr/p/prompts3ia.html
Gemini, can you respond to prompt 1?

Wait for the response.

Then enter the next prompt:
Gemini, can you respond to prompts 2 and 3?

The benefit is twofold: it maintains a version-controlled record of instructions (on GitHub or a server),
and it avoids having to open or copy-paste dense pages into the chat.
It’s essentially a "Remote Procedure Call" (RPC) for prompts.

im2.png