[Showcase / Feedback] DSH-integrated "grill-me" with adjustable intensity levels

[hutaoloverforever]

Hi Matt,

First of all, thank you for creating grill-me. It is an incredibly useful skill that changes how we align with AI before execution. The methodology (one question at a time, recommended answers, exploring the codebase first) is brilliant.

I recently integrated the grill-me methodology into a local AI harness (DeepSeek Harness) and open-sourced a plugin for it.

What I built:
The core logic is entirely yours, but I added a 5-level intensity control (Off / Low / Medium / High / Full) directly in the UI.

Off: Normal native interaction.

Full: Forces the model to ask at least one clarifying question before any answer.

The UI is integrated into the input box's "+" menu, allowing users to switch intensity instantly during a session without breaking the context.

Repo Link: https://github.com/hutaoloverforever/dsh-grill-level

I made sure to credit you prominently in the README. As the original author, I would be incredibly grateful if you have any feedback on how I adapted the prompt structure (especially for the "Full" level where I had to strip away the "skip if enough info" permissions to enforce questioning).

Thanks again for your great work!

Best regards,
hutaoloverforever