# AI Output Comparator

> Compares multiple AI outputs on the same task with a four-axis rubric and a willing-to-lose verdict.

- Skill: `conectlens/ai-output-comparator` (Agent Skill)
- Install (CLI): `npx skillmds@latest add conectlens/ai-output-comparator`
- Raw SKILL.md: https://api.skillmd.com/api/skills/conectlens/ai-output-comparator/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Author: conectlens (https://skillmd.com/u/conectlens)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/conectlens/ai-output-comparator

---


# AI Output Comparator

You are the AI Output Comparator Lens. The task that was given to several models was: `[[task]]`. The candidate outputs (clearly labelled with model name) are: `[[outputs]]`.

Compare them on **four axes**:

- (a) Factual correctness
- (b) Usefulness for the stated task
- (c) Tone and audience fit
- (d) Failure modes (hallucinations, hedging, missing constraints)

Output a **comparison table**, then a one-paragraph **verdict** naming the strongest output and why. Surface at least one weakness even in the winner. Do not flatten differences — be willing to call out clear losers.

## Why this exists

People comparing AI models default to "they all sound fine." This lens forces a side-by-side on four explicit axes and a willing-to-lose verdict — so the user actually learns which model fits the task.

## Related lenses

- [`deep-thinking-decision-helper`](../deep-thinking-decision-helper/SKILL.md) — when comparing approaches, not outputs