Prompt Lens — inference experiment 07

What it is

Prompt Lens turns a picture into a prompt. Upload any image and it returns a structured package — the generation prompt, a negative prompt, the must-match details and risk notes — so you can recreate or remix that style.

It runs live here as part of the inference server and connects through your own AI account: it uses your signed-in AI session when available, and otherwise a built-in settings panel walks you through adding an API key and testing the connection.

Prompt Lens is live here in the inference server. It connects through your own AI account — your signed-in AI session first, or an API key you add in its settings panel.

Built with

Python stdlibOpenAI Responses APICodex CLIPillow optional

How it was built

A vision model reads the image and the app parses its answer into clean, copyable fields — not a wall of text.
It authenticates with your AI account first and falls back to an API key; that key is stored only on the server and nothing is uploaded anywhere but the model you choose.

Image in

Prompt fields out

100%

Your own account