aif-grounded

To enforce evidence-based verification, you need a reliability gate that mandates evidence collection and tracks unknowns before answering. This Skill applies a four-step workflow to verify evidence and ensure responses are fully grounded.

Handling uncertainty requires explicit unknowns tracking and a clear confidence gate. This Skill enforces a 0–100 confidence scale to decide whether to answer or return a missing information checklist, preventing guessing in high-stakes domains.

To get a missing information checklist when confidence is below 100%, use an evidence-based reliability gate. This Skill outputs a clear checklist detailing exactly what evidence is missing instead of providing an unverified answer.

Yes, you can use evidence-based reasoning for tasks with changeable facts in high-stakes domains. This Skill is specifically designed to apply mandatory verification, ensuring grounded responses where facts frequently change.

You should not use a grounded AI approach when a strict 100% confidence gate is unnecessary or when missing information checklists would hinder low-stakes tasks. Grounded verification is designed for high-stakes domains requiring strict evidence.