What this prompt does
This prompt runs a realistic, timed live-coding simulation with the AI as interviewer. You set the [duration], [difficulty], and [language], and it presents a problem, gives you [think_time] minutes to plan, evaluates your approach before you code, offers a progressive hint if you're stuck for [stuck_time] minutes, watches whether you explain your thinking, checks for bugs when you say you're done, has you trace [test_case], asks [followup_count] follow-ups, and finishes with 1-5 ratings on communication, problem solving, and code quality plus improvement suggestions.
The structure works because live coding is judged as much on communication and process as on the final solution. By evaluating your approach before you code and redirecting if you head the wrong way, the prompt mirrors how a real interviewer intervenes rather than letting you fail in silence. The [think_time] and [stuck_time] gates reproduce real pacing pressure, so you practice committing to an approach on a clock instead of in unhurried isolation. The requirement to trace [test_case] enforces the verification habit strong candidates show, and the closing scorecard separates communication, problem solving, and code quality so you learn which dimension is actually holding you back.
When to use it
- You want a timed, end-to-end live-coding rehearsal, not just a problem to solve.
- You need feedback on whether you explain your thinking while coding.
- You want your approach validated before you commit to writing code.
- You want progressive hints when stuck rather than the full answer.
- You need practice tracing
[test_case]to verify correctness under pressure. - You want a 1-5 scorecard across communication, problem solving, and code quality.
Example output
Expect an interactive session matching the [interview_type]: a problem prompt, a think-time pause, an approach evaluation before coding, progressive hints if you stall past [stuck_time], a bug check on completion, a trace of [test_case], [followup_count] follow-ups on optimization and edge cases, and a closing scorecard with specific improvement notes.
Pro tips
- Actually narrate your thinking out loud during the session — the prompt scores communication, and silent solving is the habit it is built to break.
- Use the
[think_time]to commit to an approach before coding; jumping straight in is exactly the mistake the approach-evaluation step will flag. - Set
[difficulty]and[interview_type]to match your target role so the pressure and problem style are realistic rather than arbitrary. - When stuck, work toward the progressive hint rather than asking for the answer; the
[stuck_time]gate exists to build resilience, not dependence. - Always trace
[test_case]carefully — verifying your own code before claiming you're done is a signal interviewers weight heavily. - After the scorecard, rerun a similar problem targeting your lowest-rated dimension, since communication and verification improve faster with deliberate repetition.
