The most useful Claude prompts for interview practice live inside a Claude project: upload your résumé and the job description once, put the interviewer rules and a scoring rubric in the project instructions, then open a new chat for each mock round and answer by typing or in voice mode. Below are 11 example prompts, a self-scoring rubric and a five-round mock loop, with Claude's features checked in October 2026.
Claude is Anthropic's AI assistant. Its projects keep your materials and rules in one place, so each round starts with the same context instead of a fresh paste.
What Claude offers for interview practice (as of October 2026)
| Feature | Free plan | Pro plan |
|---|---|---|
| Price | $0 | $17 a month billed annually ($200 up front), or $20 billed monthly |
| Projects (files plus instructions) | Up to 5 projects | Included |
| Voice mode (beta) | Yes: mobile, desktop app and web | Yes: mobile, desktop app and web |
| Usage limits | Reset on a rolling five-hour window | At least 5 times Free's usage per five-hour window, plus weekly limits |
Sources, checked October 2026: Anthropic's Claude plans and pricing, what are projects, use voice mode and upload files to Claude. Claude reads PDF, DOCX and TXT files, among others, and files added to a project stay available across that project's chats. Anthropic's voice mode page lists practicing for interviews among its uses, and voice conversations count toward your plan's usage limits. Max plans start at $100 a month.
Set up an interview-practice project
- Create a project per job. Name it after the role and company, such as "[Company], [role] interview."
- Add files. Upload your résumé, the job description and any notes on the company's published values to the project's files.
- Paste the project instructions (prompt 1 below). They apply to every chat in the project.
- Start a new chat for each round. A fresh chat keeps each round focused and its feedback easy to find.
- Go to voice when you're ready to speak. On the web or the desktop app, select the sound wave icon in the lower right of the chat window; on mobile, tap the sound wave icon next to the microphone. Anthropic says you can switch between text and voice in the same conversation and your context carries over.
- End and review. Select Stop on the web or desktop to leave voice mode. Text transcripts of voice conversations are saved in your chat history, so you can type a review prompt right after.
11 Claude prompts for interview practice and mock loops
These are example prompts we wrote for this guide. Replace anything in [square brackets]. Prompt 1 goes into the project instructions; the rest go into chats inside the project.
1. Project instructions: the interviewer charter
You are my interviewer and interview coach for the [role] position at [company]. Use the résumé and job description in this project's files. Rules for every chat: ask one question at a time and wait for my full answer. Never write an answer for me unless I type "model answer." Ask at least one follow-up per question about what I personally did. Before you score an answer, ask me to score myself on the rubric. Then give your scores side by side with mine, quote the words behind each score, and explain any gap of 2 points or more. Rubric, each 1 to 4: Structure (clear situation, action, result), Specificity (names, numbers, decisions), Ownership (my actions, not the team's), Result (a measurable outcome or a clear lesson). Keep feedback to five sentences per answer.
The self-score step is what makes this different from a standard mock. Comparing your score with Claude's shows where your sense of your own answers is off.
2. Plan a five-round mock loop
Using the job description, design a five-round mock interview loop for this role: a recruiter screen, a hiring manager round, a technical or skills round, a behavioral round and a final round on values and motivation. For each round, give its goal, three likely questions and what would count as a strong signal. Show it as a table. Don't answer the questions.
Run one round per chat, one a day, and you have a full loop in a week. If you're unsure what each stage is for, see what a final round interview is.
3. Audit your story inventory
From my résumé, list the eight experiences most likely to come up in a behavioral interview for this role. For each, tell me which skill in the job description it could prove and what detail is missing from the résumé that I'd need to say out loud. Then list the skills that have no story behind them.
4. Behavioral round with self-scoring
Start the behavioral round. Ask four questions that test the skills this job description stresses most. Follow the project rules: one question at a time, a follow-up on my personal actions, then my self-score before yours.
Strong stories follow a clear structure; what is the STAR method explains it, and the STAR story builder helps you draft one.
5. Deep dive on one résumé project
Pick the project on my résumé that a [hiring manager or senior engineer] would find most interesting. Ask me to explain it in two minutes. Then ask, one at a time: what the hardest decision was, what I'd change now, how I measured success and what my teammates would say I contributed. Score me on the rubric afterward.
6. Typed coding round
Act as a coding interviewer. Give me one [easy, medium or hard] problem on [topic] in [language], with examples and constraints but no hints. Let me ask clarifying questions. When I paste my solution, check correctness on two edge cases you choose, state the time and space complexity, and ask one follow-up that changes a constraint. Don't show your solution unless I type "model answer."
Write your code in your own editor, not in the chat, so you practice the way you'll be tested. The coding interview cheat sheet lists the patterns worth drilling first.
7. Work-sample round
Give me a realistic task someone in this role would face in their first month, based on the job description, such as prioritizing a backlog, reviewing a plan or diagnosing a drop in a metric. Describe it in under 120 words. Let me talk it through, ask me two probing questions, then tell me what a strong hire would have done differently.
8. Voice warm-up
We're in voice mode. Ask me five short questions for this role, one at a time, in one sentence each. I'll use push-to-talk. After each answer, say only "clear" or "unclear," and save the details for when I say "debrief."
9. Redo the weakest answer
Look at the scores in this chat and find my lowest-scoring answer. Ask me that question again. After my new answer, compare both versions on the rubric and tell me exactly what improved and what didn't.
10. Hiring committee debrief
Here are my scorecards from the last five rounds: [paste]. Act as a hiring committee reviewing them. Summarize the strongest signals and the biggest concerns, decide whether you'd move me forward and why, and name the one skill to fix before the real interview.
Don't count on one chat knowing what happened in another. Paste each round's scores into a single "scorecards" document and add it to the project files, so later rounds and this debrief can see your history.
11. Track progress in an artifact
Create a simple table as an artifact to track my mock interview practice: columns for date, round, question, my self-score, your score and the fix to try next. Fill in today's round from this chat. I'll paste future results so you can update it.
Score yourself with a rubric
Here is the rubric from prompt 1 with anchors, an example we wrote for this guide. Score yourself first, then compare with Claude's score.
| Criterion | A 4 sounds like | A 1 sounds like |
|---|---|---|
| Structure | Context in one or two sentences, then your actions, then the result | Background that never reaches a point |
| Specificity | Names a tool, a number, a deadline or a decision you made | "We worked hard and it went well" |
| Ownership | "I proposed," "I decided," "I convinced" | "We," "the team," "it was decided" |
| Result | A measurable outcome, or a clear lesson you applied later | Ends with the task, not what changed |
A gap of 2 or more points between your score and Claude's is the most useful signal you'll get. If you rate yourself higher, the answer may sound better in your head than out loud. If you rate yourself lower, you may be underselling work that's actually strong. For how scored feedback works in dedicated tools, see what a score out of 10 means.
Limits of Claude for interview practice
- Voice mode is in beta. Anthropic says it works best on a phone, and languages other than English are also in beta.
- Usage limits. Limits reset on a rolling five-hour window, and voice conversations count toward them. Long voice rounds on the free plan can hit the limit.
- Five projects on Free. If you're interviewing at several companies at once, reuse one general project and add each job description as a file.
- No clock. Claude doesn't time your answers. Run a timer and keep behavioral answers to about two minutes.
- Scores are relative. A rubric makes feedback steadier, but scores can still shift with wording and chat length. Compare trends, not single numbers.
- Company details can be out of date. Check interview formats on the company's careers pages; our guide to researching a company with AI covers source-checking.
- Privacy. Don't upload a current employer's confidential documents to any assistant.
When a dedicated scored mock interview helps
Claude is a strong coach for planning a loop, auditing stories and reviewing answers in depth. A dedicated mock interview helps when you want spoken rounds with the same scoring every time and no setup. CoPilot Interview's Mock Interview, in the desktop app for Windows and macOS, builds the interview from your résumé and the job description. You choose the level, 3, 5 or 8 questions, the mix of behavioral, technical, system design and coding questions, your coding language and an optional focus. It reads each question aloud, scores each spoken answer out of 10 with what to fix and a model answer, and ends with a report card.
| Task | CoPilot Interview Mock Interview | Claude with these prompts |
|---|---|---|
| Spoken rounds with consistent scores | Best choice: questions read aloud, every answer scored out of 10 with what to fix and a model answer | Voice mode (beta) plus the rubric you write |
| Self-scoring and calibration | Scores and a report card from the app | Strong: you can build a self-score step into every round |
| Planning a loop and auditing stories | Set the mix and focus for each session | Strong: tables, debriefs and artifacts on request |
| Cost to start | Free: 15 minutes a day, no credit card | Free plan with usage limits; Pro from $17 a month billed annually |
Disclosure: we make CoPilot Interview.
If you like Claude's style, CoPilot Interview's Pro+ plan ($49.99 a month) includes ChatGPT, Gemini and Claude models. To start without installing anything, try the browser version of the free AI mock interview, with 113 real questions across 9 tracks. For spoken practice in other assistants, see our guides to Gemini Live and ChatGPT voice mode. Rehearse before the interview, not during it. Follow the rules of the employer or assessment you're in — many people use CoPilot Interview only to practice, which works on the free plan.
Spoken mock interviews with scores built in
CoPilot Interview's Mock Interview reads each question aloud, listens to your answer and scores it out of 10 with what to fix and a model answer. Start in your browser, or use the free app with 15 minutes a day and no credit card.
Try a free AI mock interviewFAQ
Can Claude run a mock interview by voice?
Yes. As of October 2026, voice mode is a beta feature on all Claude plans, on mobile, the desktop app and the web. Start a chat, select the sound wave icon and answer out loud. Your context carries over when you switch between text and voice.
Is Claude free for interview practice?
Yes, with limits. The free plan includes voice mode and up to five projects, and usage resets on a rolling five-hour window. Pro costs $17 a month billed annually or $20 monthly.
What should go in Claude project instructions for interview prep?
Your interviewer role, the rule to ask one question at a time and wait, a ban on writing answers for you, a follow-up rule and a scoring rubric. Prompt 1 above includes all five.
How do I stop Claude from answering the questions for me?
Say so in the project instructions: never write an answer unless I type a trigger phrase such as "model answer." If it slips in a long chat, restate the rule or start a new chat in the project.
Can Claude score my interview answers accurately?
It can score them consistently enough to be useful if you give it a rubric with anchors and ask for quoted evidence. Treat scores as a trend across sessions, not a precise measurement.
Should I use Claude or ChatGPT for interview practice?
Either works with good prompts. Claude's projects suit a structured loop with a fixed rubric, while ChatGPT's voice mode has fixed allowances per rolling 24 hours on paid plans. You can also use one for planning and the other for spoken reps.