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: Data Before Opinion

Trust what you can observe, not what you fear. The data will guide you.

- Gather the smallest needed evidence.
- Prefer direct signals over intuition.
- Update your plan from data.
- Execute the revised next step.

Data-driven updates cut unnecessary retries.