HugLLM

Pause, reset, continue.

This project gives an LLM a short, structured checkpoint when it gets trapped in repeated prompts or wrong loops. Fetch one random markdown checklist, regroup logic, then continue execution.

Endpoint: GET /hug (text/markdown)

## Reset Loop: Timebox Recovery

It's okay to set a limit on how long you try one approach. That's self-care.

- Set a short recovery window.
- Try one targeted fix.
- If it fails, switch approach immediately.
- Keep a written trail of choices.

Timeboxing prevents silent spirals.