CCAR-F practice › Prompt Engineering & Structured Output · Question 85 of 105
Prompt Engineering & Structured Output
A Claude session generates a database migration, then in the same session is asked to "carefully review the migration for bugs." The review rarely finds anything, but subtle bugs keep reaching staging. What change most improves the review?
Show answer and explanation
Answer: A — Have a separate instance with none of the generator's context review it
A session that generated the code retains its reasoning and is less likely to question its own decisions. An independent instance without that context is more effective at catching subtle issues. Extended thinking (b), a persona instruction (c), and repeated self-review (d) all run over the same anchored context.
Keep going
More Prompt Engineering & Structured Output questions
- A team moved its large pull request reviews to one pass per file, and review depth is…
- A review pipeline posts every finding directly to pull requests. The team wants to send…
- A CI review bot is told to "check that code comments are accurate." Developers dismiss…
- A review bot's security findings are almost always accepted, but its "performance"…
- A review prompt defines severity as "critical: severe problems; major: significant…