@quanta.cortex: Claude says it’s done. Did the tests run? Ask to see the results. If you’re moving a Claude Code workflow to Fable 5.2, start with the prompt you already use. Give the task a finish line and ask to see the test results. For a small fix, say which files or feature Claude can change. For a longer job, ask it to check in after a milestone you care about. The carousel covers ten adjustments, from choosing effort to controlling reply length. The last slide gives you a prompt to try on a small bug you can verify yourself. The FrontierCode 2.0 figures are from Anthropic’s benchmark. Your results and costs depend on the task and setup. Anthropic’s prompting guide https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-2 Anthropic’s release and benchmark https://www.anthropic.com/claude-fable-and-mythos-5-2 #ClaudeFable #ClaudeCode #PromptEngineering #AICoding #RAG