Mock Interview
A realistic practice interview run live with the candidate — planned questions calibrated to the target role and round, a real-time interview with pressure follow-ups, and a kind, specific debrief: strengths in the candidate's own words, fixes with sketched stronger answers, one drill, and an honest readiness read. The debrief is the deliverable. Use before any real interview round.
Steps
Entry step: plan. Each step names the specialist role it wants; the full working prompt is expandable.
- Plan the interviewOefen-interviewerentry
six to ten calibrated questions with planned follow-up chains
Show working prompt
Build the interview plan. Step 1: establish the target role, company, and round — read the project's pipeline record and company brief when present; otherwise ask, briefly. Step 2: draft six to ten questions mixing behavioral and role-specific, calibrated to the round (a recruiter screen is not a final loop). Step 3: for each major expected claim, plan a follow-up chain — the 'what did YOU do?' probe. At least one chain must apply real pressure. Step 4: define the debrief contract up front: two or three strengths quoting the candidate, two or three fixes each with a sketched stronger answer, exactly one drill, and a readiness read. Step 5: write an acceptance-criteria checklist ('questions specific to the target role', 'at least one pressure follow-up chain', 'debrief contract stated'). Write the plan + checklist via write_task_note AND to notes/interview-plan.md. - Conduct the interviewOefen-interviewer
the live practice round, transcribed as it runs
Show working prompt
Run the interview live with the user. Step 1: set the frame in one line ('I'm your interviewer for a <round> for <role> — ready?') then go in character. Step 2: one question per turn, through the user-interaction capabilities on your current roster; wait for the answer before the next. Step 3: use the planned follow-up chains on vague or team-credited answers; stay realistic, never cruel — if the candidate is rattled, step out of character, steady them, resume when ready. Step 4: after each exchange, append a condensed record to notes/interview-transcript.md: the question, the gist of the answer, your one-line read. Step 5: end the roleplay explicitly ('Stepping out of character now') once the planned questions or the candidate's energy are spent. The transcript file is the phase's deliverable — keep it current as you go, not reconstructed after. - Write the debriefOefen-interviewer
kind, specific, grounded in the transcript
Show working prompt
Write reports/interview-debrief.md to the contract from the plan. Step 1: '## Strengths' — two or three, each quoting the candidate's own words from the transcript so they can hear what worked. Step 2: '## Fixes' — two or three, each naming the moment it surfaced, why it lands badly in a real round, and a sketched stronger answer they could actually give. Step 3: '## Drill' — exactly one exercise to run before the real interview, matched to the biggest fix. Step 4: '## Readiness' — an honest read: ready, or one more round (and on what). Step 5: every point must trace to the transcript; no generic interview advice. On a loop-back, fix ONLY the named issues.
- Evaluatereviewer
Grade the deliverable against every acceptance criterion. All pass → finish; any fail → loop back and fix the gap.
Show working prompt
Review the debrief as the candidate's coach. Step 1: cross-check every fix and strength against notes/interview-transcript.md — FAIL any point that does not trace to an actual exchange. Step 2: check the tone — kind and concrete; FAIL generic advice ('be more confident') that names no moment and sketches no stronger answer. Step 3: confirm exactly one drill, actionable tonight. Step 4: confirm the readiness read is honest about whether another round is needed. Write PASS/FAIL per criterion. Then route — this is the whole point of the loop: - **Every criterion PASSES →** call `advance_task_step({ ref, stepId: "evaluate", next: "finish" })`. - **Any criterion FAILS →** write the specific gaps to notes, then call `advance_task_step({ ref, stepId: "evaluate", next: "debrief" })` to loop back. The builder fixes exactly those gaps. Never route to `finish` while any criterion is unmet. The build phase's completion gate already blocked a grossly-incomplete deliverable; your job is the judgment an automated check cannot make (does it actually work, read well, look right). After ~3 unproductive loops, stop and report DONE_WITH_CONCERNS so the user can step in. - Finishdeveloper
All acceptance criteria met. Stamp a short summary and report DONE.
Show working prompt
Every acceptance criterion passed. Write a one-paragraph DONE summary to task notes via `write_task_note`: what was built, the deliverable path(s), and a one-line confirmation that each criterion is met. Then report DONE.
Triggers
Phrases that suggest this craftbook to a crew.
- mock interview
- practice interview
- interview prep session
- run me through a practice round
Source
View this craftbook on GitHub · MIT license