AI Truth Mirror: AI Output Authenticity Checker

Problem

AI-generated answers often look polished while omitting source, timestamp, units, scope, or assumptions, or while turning vague claims into definitive statements. When reviewing API notes, pricing, statistics, or professional conclusions, the immediate need is not another rewrite, but a quick split between content that can enter a decision path and content that must go back for manual verification.

Workflow

The skill checks an existing AI output across five dimensions:
- Completeness: whether data source, time, units, applicable scope, and assumptions are present.
- Logical consistency: contradictions, calculation errors, concept drift, or extreme wording.
- Timeliness: whether time-sensitive values such as price, weather, or statistics are dated and current.
- Source credibility: whether the claim cites official docs, reports, or papers instead of vague phrases.
- Domain fit: whether the topic is within a reliable judgment range, with an explicit flag when it is not.

It then emits an A-D reliability report that separates credible, questionable, and clearly problematic parts, making follow-up review easier.

Boundary

It does not replace primary documents, audit systems, domain experts, or live data verification. Output quality depends on the context supplied by the user; for highly specialized or fast-moving topics, treat the report as a review checklist rather than a final factual verdict.