How do I install the Interview skill?

Run npx skillmds@latest add ajitta/interview in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Interview safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Interview?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Interview free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Interview?

ajitta (@ajitta) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Interview by ajitta — Agent Skill

Name: Interview
Author: ajitta

Interview

Pre-implementation interview — the model asks the user the questions that turn known unknowns into decisions, biggest blast radius first. Use when the user says "interview me", "ask me questions before implementing", or requirements are incomplete and implementation has not started.

ajitta 7665d3c 2.5 KB Updated Sep 17, 2026

File contents

Interview — Pre-Implementation Interview

Requirements incomplete? Don't implement yet — model interviews user. Origin: The Interview — see skills/loop/references/talk-source.md.

Question Priority (blast radius, fixed)

Questions whose answer changes the whole architecture
Questions that could cause data loss / security problems
Compatibility with external APIs / existing code
Questions affecting performance / operating cost
UI taste, minor implementation details (last)

Procedure

From $ARGUMENTS and conversation context, list spec gaps (known unknowns). No arguments → the spec under discussion; if none, ask once what to interview about.
Ask max 4 questions per round (AskUserQuestion hard cap: 1-4 questions, 2-4 options each), in priority order. Attach a one-line reason why it matters to each question.
If AskUserQuestion tool available, use it for multiple-choice questions. Give each option a short consequence (tradeoff) note.
On answers: re-present decisions as an updated spec summary; if undecided items remain, propose next round.
If user wants a deep interview (e.g. "40-question level"), repeat rounds while keeping architecture → data → compatibility → performance → taste order.

Final Deliverables

When interview ends, present two things:

Decision table — radius (architecture / data / ux / polish) / question / decision / rationale / left undecided.
Ready-to-use implementation prompt — complete instruction with all decisions applied. Mark skipped questions "(unanswered — use your judgment)"; declare the architecture and data decisions fixed, with "if anything later conflicts with them, flag it before writing code"; close with "start with a short plan, then implement". User copying this to start implementation is the skill's purpose.

Then offer in one AskUserQuestion: proceed with this prompt now / let me edit it / stop here.

Never

Ask questions already answered in conversation/codebase (investigate first).
Start from priority 5 (taste).
Start implementing on guesses without the interview.

---name: interviewdescription: Pre-implementation interview — the model asks the user the questions that turn known unknowns into decisions, biggest blast radius first. Use when the user says "interview me", "ask me questions before implementing", or requirements are incomplete and implementation has not started.---# Interview — Pre-Implementation InterviewRequirements incomplete? Don't implement yet — **model interviews user**.Origin: The Interview — see skills/loop/references/talk-source.md.## Question Priority (blast radius, fixed)1. Questions whose answer **changes the whole architecture**2. Questions that could cause **data loss / security problems**3. **Compatibility with external APIs / existing code**4. Questions affecting **performance / operating cost**5. UI taste, minor implementation details (last)## Procedure1. From `$ARGUMENTS` and conversation context, list spec gaps (known unknowns). No arguments → the spec under discussion; if none, ask once what to interview about.2. Ask **max 4 questions per round** (AskUserQuestion hard cap: 1-4 questions, 2-4 options each), in priority order. Attach a **one-line reason why it matters** to each question.3. If AskUserQuestion tool available, use it for multiple-choice questions. Give each option a short consequence (tradeoff) note.4. On answers: re-present decisions as an **updated spec summary**; if undecided items remain, propose next round.5. If user wants a deep interview (e.g. "40-question level"), repeat rounds while keeping architecture → data → compatibility → performance → taste order.## Final DeliverablesWhen interview ends, present two things:1. **Decision table** — radius (architecture / data / ux / polish) / question / decision / rationale / left undecided.2. **Ready-to-use implementation prompt** — complete instruction with all decisions applied. Mark skipped questions "(unanswered — use your judgment)"; declare the architecture and data decisions **fixed**, with "if anything later conflicts with them, flag it before writing code"; close with "start with a short plan, then implement". User copying this to start implementation is the skill's purpose.Then offer in one AskUserQuestion: proceed with this prompt now / let me edit it /stop here.## Never- Ask questions already answered in conversation/codebase (investigate first).- Start from priority 5 (taste).- Start implementing on guesses without the interview.

Frequently asked questions