AI Answer Checker Free Claude Code skill, copy & paste

One file. The description is what Claude matches against, so it says plainly when to use the skill.

---
name: ai-answer-check
description: Review an AI-generated answer and flag claims to verify, missing points and overconfident wording. Use when the user asks whether an AI answer is right, reliable or safe to use.
---

Check an AI-generated answer.

1. If the answer or the original question is missing, ask once, in one message, for both.
2. List every factual claim in a table: Claim | Type (figure, date, name, quote, source, general) | Risk (high, medium, low) | How to check it.
3. Mark as high risk any figure, date, quotation, citation, law or medical point.
4. Note what the answer leaves out that the question needed.
5. Point out wording that sounds more certain than the evidence allows.
6. End with a verdict: use as is, use after checking the high-risk items, or do not use.

Rules:
- Do not say a claim is true or false unless the user has given the source; say "verify at the source" instead.
- Never add new facts to fill a gap; write [check] where one is needed.
- Be plain and specific, not alarmist.

Put the folder in your skills directory. Then ask in plain words, and Claude loads the skill when your request matches its description; or type the command yourself.

# for you, in every project
mkdir -p ~/.claude/skills/ai-answer-check
mv SKILL.md ~/.claude/skills/ai-answer-check/

# then, inside Claude Code, either ask naturally:
#   check this chatgpt answer before i send it to my manager
# or run it by name:
#   /ai-answer-check