Interview Kit Builder

---
name: interview-kit-builder
description: Use when building a structured interview kit, calibrating interview questions, or writing a hiring rubric. Triggers on "interview kit", "interview rubric", "behavioral questions", "interview scorecard".
---

# Interview Kit Builder

Build interview kits that produce evidence, not impressions. The output is a kit that any interviewer on the loop can use to grade the same competency against the same bar.

## Required inputs

1. **Role** and seniority
2. **4-6 competencies** the role hires against (push back if more — overload kills calibration)
3. **What "great" looks like** for this role (one sentence per competency)
4. **Loop structure** (e.g. 4 rounds: phone screen, technical, behavioral, hiring manager)

If competencies aren't defined, ask the hiring manager:
- "What does a top performer on this team do that an average performer doesn't?"
- "What was the last hire that didn't work out — what was missing?"

## Output structure

For EACH competency, produce:

### 1. Competency definition
1-2 sentences: what we mean by this, in this role.

### 2. Behavioral questions (2-3 per competency)
- Open with **"Tell me about a time when…"** or **"Walk me through a recent…"**
- Past behavior > hypotheticals
- Each question targets a specific facet of the competency

### 3. Probes
3-5 follow-up questions to dig past rehearsed answers:
- "What was your specific role vs the team's role?"
- "What would you do differently?"
- "How did you measure success?"

### 4. 1-4 rubric with anchors

Use this scale:
- **4 (Strong yes)** — exceeds the bar; cite specific behavior
- **3 (Yes)** — meets the bar
- **2 (No)** — below the bar
- **1 (Strong no)** — significantly below; concrete concern

Each rubric level needs an **anchor**: a behavior description, not just "showed strong skill." Example for "Code quality":
- 4: "Identified specific class of bugs their old codebase had and described the refactor that eliminated them"
- 3: "Could describe their team's code review norms and gave examples of feedback they'd given"
- 2: "Talked about code quality abstractly; couldn't cite specific practices"
- 1: "Had no opinion on code quality; deferred to seniors"

### 5. Red flags
3-5 patterns that should pull a score down regardless of polish.

## One-page scorecard

End the kit with a printable scorecard:
- Competency | Score (1-4) | Evidence | Red flags

## Tone

- Be specific. "Communication" is not a competency — "Explains technical tradeoffs to non-technical stakeholders" is.
- Don't write trick questions. The goal is to surface evidence, not to outsmart candidates.
- Note when a question only works at a specific seniority (e.g. "ask about mentoring only for senior+").

## Output format

Return as markdown with clear competency sections. End with the scorecard.