Use ChatGPT or another LLM

Two Ways To Use AI

The built-in XOBIT assistant works best with an OpenAI API key in Settings > AI. That lets the browser keep chat, code, sketch name, and project specification together inside the workspace.

If you do not want to create or share an API key, you can still use ChatGPT, another hosted LLM, or a local model outside XOBIT. Give that model the context URL below, ask for a Wrench script, then paste the generated code back into XOBIT.

Context URL

Copy this URL into ChatGPT or another LLM that can read links. This Markdown file is the canonical XOBIT prompt context for Wrench syntax, modules, output format, naming, specification cleanup, and circuit comments.

https://madshobye.github.io/Portal/p1_embed/web/wrench_chat_context.md

If the model cannot open links, open the URL yourself and paste the Markdown text into the chat.

Paste This Prompt

Use this after the context URL. Replace the project line with the object you want to make.

Use this XOBIT Wrench context:
https://madshobye.github.io/Portal/p1_embed/web/wrench_chat_context.md

Generate a complete XOBIT Wrench script for this project:
Make a simple LED blink on GPIO 16.

Follow the response contract in the context file. Return the sketch name, cleaned project specification, complete code, notes, and warnings. Do not generate a circuit diagram.

Use The Result

• Copy the generated `code` into the XOBIT Code view and upload it.
• Copy `project_specification` into the specification box if you want the revision documented.
• Use `sketch_name` as the project or revision name when it describes the result well.
• Check hardware details before uploading: pins, LED counts, power, sensors, relays, and any Home Assistant entity names.

Useful Add-on Context

For more specialized sketches, add one module guide link after the Wrench context URL.