CCA-F: Claude Certified Architect — Foundations

By Anthropic · Question Mode

QuestionQ129

Prompt Engineering & Structured Output

You are integrating Claude Code into your Continuous Integration/Continuous Deployment (CI/CD) pipeline. The system performs automated code reviews, creates test cases, and gives feedback on pull requests. You must design prompts that deliver actionable feedback while minimizing false positives.

Your automated reviewer uses one prompt for security issues, API design, and business-logic correctness. Your evaluation suite shows strong recall for API-design findings (82%) but weak recall for business-logic edge cases in quiz scoring (34%). After you add few-shot examples of logic bugs to the prompt, logic recall rises to 41% but API-design recall falls to 68%.

How should you address this trade-off to improve detection across both categories?

Explanation

Independent review dimensions benefit from focused evaluation prompts with their own task-specific instructions and representative examples. Running separate security/API-design and business-logic reviews prevents new business-logic examples from displacing attention from API design; consolidating their outputs preserves a single pull-request feedback workflow. Anthropic’s evaluation guidance supports comparing prompt variants against test cases to identify and iterate on such performance differences.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!