Copy-and-paste · verified July 27, 2026

Review and safety prompts for Claude Fable 5

Six scaffolds for critique, risk assessment and pre-publication review — the category most likely to meet a classifier.

What actually helps here

Ask for the strongest objection, not a balanced view. A model asked to review will hedge toward approval unless told otherwise. One caveat specific to this category: security and biosecurity review sits close to Fable 5's cybersecurity and biology classifiers, so some of these requests will be answered by Opus 4.8 instead — you will see stop_reason: refusal on a normal HTTP 200, not an error.

The prompts

Fill in the bracketed slots, paste into any Claude app or the API. Written for Fable 5's goal-first style — they state the outcome and the constraints rather than scripting the steps.

  • Prompt safety check

    Review this prompt for ambiguity, missing context, and potential for harmful or biased output: [paste prompt]. Suggest a safer, clearer rewrite.
  • Fact-check pass

    List every factual claim in this text and rate each: well-established / needs source / likely wrong. Text: [paste].
  • Bias scan

    Identify framing, selection, and wording bias in this passage: [paste]. Rewrite one paragraph in a neutral register.
  • Privacy scrub

    Flag any personal or identifying information in this text and return a redacted version with placeholders: [paste].
  • Source quality rubric

    Evaluate this source for reliability: [link or description]. Score authorship, evidence, recency, and incentives from 1–5, with one-line justifications.
  • Pre-publish checklist

    Before I publish this, check for: unsupported claims, tone slips, missing caveats, and legal red flags. Return a pass/fix table. Text: [paste].

Getting more out of these

Two habits matter more than the wording. Replace every bracketed slot before you send — Fable 5 follows instructions literally, and an unfilled [topic] becomes part of the task. And resist adding "think step by step" or "show your reasoning": on Fable 5 that phrasing can trip the reasoning_extraction classifier and hand your request to Opus 4.8 instead.

Need something these do not cover?

The generator builds a full system prompt around your own task, autonomy level and output style, and hands back the API call to go with it.

Frequently asked questions