Prompting Claude Fable 5.2
Web page · https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-2
First seen by Oriel on .
AI summary
This guide explains Claude Fable 5.2 prompting patterns for effort, task completion, code ownership, progress updates, delegation, approvals, migration, reports, and refusals.
- Start at high effort, then validate lower settings; high provides most capability gains, while xhigh and max increase tokens and time for uneven benefits.
- For long agentic tasks, instruct Claude Fable 5.2 to diagnose ordinary errors and continue, but retain confirmation steps for risky or irreversible actions.
- Describing a codebase as the model’s own increased duplicated-code consolidation from under 10 percent to about 75 percent, while also increasing output and unsolicited changes.
- Progress updates arrive as thinking blocks and are empty when thinking.display is "omitted"; use "updates" or "summarized" and render them between tool calls.
- Review earlier prompts and harnesses because delegation instructions, visible token budgets, and edits to prior turns can increase cost, side effects, or invalidated thinking blocks.
#claude fable #prompt engineering #agentic coding #anthropic api
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hacker News: 9, latest observation ; +5 over observed period since
Comments
Comment count reported by the source.
- Hacker News: 0, latest observation ; 0 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hacker News | 50453513 | https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-2 Read discussion |