A Claude Code skill for software engineering interview preparation.
Structured mock interviews with per-answer scoring, session notes, spaced repetition, and progress tracking — all running locally inside your Claude Code session.

---

Contents

• Features
• Installation
• Quick Start
• Session Types
• Modes
• JD & CV Tailoring
• Mid-Session Commands
• Configuration
• Session Notes & Progress Tracking
• Customising for Your Stack
• File Structure
• Contributing

---

Features

Category	What it does
Session types	Behavioral · Technical · System Design · Mixed · Post-interview debrief
Modes	Standard (live feedback) · Pressure (debrief at end)
Tailoring	JD-based · CV-based · JD + CV combined with gap analysis
Company mode	Style adjustments for Amazon, Google, Meta, Microsoft, Apple, Stripe, Netflix
Warm-up	Optional "Tell me about yourself" opener with structure and timing feedback
Scoring	Honest per-answer scores with model answers and gap analysis
Retry	Say "let me try again" — get a delta score on your second attempt
Spaced repetition	Questions scored ≤ 5 automatically resurface as variants in future sessions
Code evaluation	Correctness · complexity · edge cases · style
Time awareness	Estimates speaking time per answer · flags answers that are too long or too brief
Session log	Auto-written after every session — tracks questions asked, scores, dates
Progress tracking	Per-topic score trends across sessions · weak area ranking · "show my progress"
Mid-session commands	Hint · skip · harder/easier · retry · "what are they testing for?" · export

---

Installation

git clone https://github.com/glass-uk/interview-prep ~/.claude/skills/interview-prep
cp ~/.claude/skills/interview-prep/references/config.example.md \
   ~/.claude/skills/interview-prep/references/config.md

Then edit references/config.md with your stack and preferences (see Configuration).

---

Quick Start

In any Claude Code session:

Interview me
Start a mock interview — Kafka focus, pressure mode
Practice behavioral questions
Give me a system design question
Post-interview debrief

The skill reads your config, scans for weak areas from previous sessions, and asks a single setup question before starting.

---

Session Types

1. Behavioral

STAR structure enforced on every answer. Missing components called out explicitly.
Scoring requires: specific situation, personal responsibility (not "we"), concrete actions, quantified result.

2. Technical

Questions drawn from your configured stack. Probing follow-up after every answer before scoring.
Pre-loaded questions for Java · Spring Boot · Kafka · MongoDB · DSA.

3. System Design

7-phase structured framework: requirements → capacity → architecture → data model → API → deep dive → failure modes.
Guided through each phase. Skipped phases prompted before scoring.

4. Mixed

Rotates types: behavioral → technical → system design → behavioral.

5. Post-Interview Debrief

Describe a real interview you just had. Get model answers and gap analysis for every question asked.
Identifies patterns across your answers ("your technical answers lack production depth") and recommends what to revise before the next round.

---

Modes

Standard — feedback, score, and model answer after each question. Best for learning.

Pressure — interviewer responds as a real interviewer would (brief acknowledgement, no feedback). Full debrief delivered at the end. Best for simulation.

---

JD & CV Tailoring

Paste either or both during session setup:

JD only — question weighting shifts toward what the role requires. Model answers framed to the role context. Company-specific style applied if the company is recognised.

CV only — questions drawn from your actual stated experience. Claimed skills held to the depth you claimed. Behavioral questions prompt you to draw from specific roles listed on your CV.

JD + CV — overlap identified (skills you claimed that the role requires) and treated as the highest-priority questions. CV gaps vs JD requirements surfaced upfront:

"The role lists Kubernetes as a requirement but it doesn't appear on your CV. Want to include a question on that?"

Company Style Adjustments

Company	What changes
Amazon	Behavioral questions weighted toward Leadership Principles. Expect STAR with metrics.
Google	Algorithmic depth required. Time/space complexity discussion expected.
Meta	Product thinking + "why this over alternatives?" follow-ups.
Microsoft	Balanced technical and behavioral. Growth mindset framing.
Apple	Technical depth and correctness emphasis.
Stripe	API design and reliability focus.
Netflix	Autonomy, ownership, and trade-off reasoning signals.

---

Mid-Session Commands

Say any of these at any point during a Standard session:

Command	What happens
give me a hint	One directional nudge. Score penalised by 1 point.
how am I doing?	Running score table: question · topic · score · session average.
skip this question	Move on. Not scored.
harder / easier	Adjusts difficulty of the next question.
what are they testing for?	Explains the intent and common traps, then re-asks.
let me try again	Second attempt on the same question. Shows Attempt 1: X → Attempt 2: Y (Δ +Z).
export session	Writes full session to notes/exports/[date]-session.md. Session continues.
save notes	Saves model answers and checklists to topic notes files.
end session	Triggers session summary and auto-writes session log.

In Pressure mode all mid-session commands are ignored — the interviewer responds: "Let's keep going."

---

Configuration

Copy the example and fill it in:

cp references/config.example.md references/config.md

tech_stack: Java, Spring Boot, Kafka, MongoDB
target_level: senior
notes_path: ~/.claude/skills/interview-prep/notes
time_tracking: on
warmup: ask

Field	Options	Description
tech_stack	comma-separated list	Technologies to practise. Used for question selection, dynamic generation, and scoring calibration.
target_level	junior · mid · senior · staff · principal	Sets difficulty framing and feedback expectations.
notes_path	any path	Where session notes are saved. Default: notes/ inside this directory.
time_tracking	on · off	Estimates speaking time per answer and flags if outside the target window.
warmup	ask · always · off	Controls whether the "Tell me about yourself" opener is included each session.

config.md is gitignored — it stays local.

---

Session Notes & Progress Tracking

Notes

Saved per topic to notes/[topic]/[Topic] - [Subtopic].md. Each note contains:

• The question (verbatim)
• Your answer summary
• The full model answer
• Key gaps
• A to-review checklist

Say save notes or end session and save to trigger saving.

Session Log

notes/session-log.md is written automatically after every session. Tracks questions asked, scores, and dates. Powers:

• Cross-session question deduplication
• Spaced repetition (questions scored ≤ 5 resurfaced as variants)
• Score trend analysis

Progress Check

Say show my progress or what have I covered? at any time to get:

• Topics practiced
• Per-topic score trends (improving / declining / stable)
• Ranked weak areas
• Recommended focus for next session

---

Customising for Your Stack

The pre-loaded question bank covers Java · Spring Boot · Kafka · MongoDB · System Design · DSA · Behavioral.

To adapt it:

1. references/config.md — set tech_stack to your technologies
2. references/question-bank.md — replace or add technology sections using the [M/H/VH] format
3. references/scoring-rubric.md — add tech-specific depth criteria (see existing Java/Kafka/MongoDB sections as examples)
4. references/notes-map.md — update the topic → subfolder mapping to match your stack

The skill generates questions dynamically when the bank runs out, using your configured stack.

---

File Structure

interview-prep/
├── SKILL.md                          # Skill definition — read by Claude Code at runtime
├── README.md                         # This file
├── .gitignore                        # Excludes config.md and notes from the repo
├── assets/
│   └── banner.svg                    # README banner
├── references/
│   ├── config.example.md             # Config template — copy to config.md and edit
│   ├── config.md                     # Your personal config (gitignored)
│   ├── question-bank.md              # Question library — customise for your stack
│   ├── scoring-rubric.md             # Scoring criteria per question type
│   ├── system-design-framework.md    # 7-phase system design framework
│   └── notes-map.md                  # Topic-to-folder mapping for saved notes
├── examples/
│   ├── behavioral-session.md         # Example: behavioral exchange with STAR feedback
│   ├── technical-session.md          # Example: technical exchange with follow-up
│   └── system-design-session.md      # Example: system design using the 7-phase framework
└── notes/                            # Your saved notes and session log (gitignored)
    ├── session-log.md                # Auto-written after every session
    ├── exports/                      # Full session exports
    ├── debrief/                      # Post-interview debrief notes
    └── [topic]/                      # Per-topic question notes

---

Contributing

PRs welcome for:

• Question bank sections — Go, Python, Rust, TypeScript, React, PostgreSQL, Redis, Kubernetes, etc.
• Scoring rubric sections — tech-specific depth criteria for other stacks
• Example sessions — additional worked examples showing the skill in action
• Company style entries — additions to the company mode lookup table
• System design framework improvements — new phases, better prompts, additional failure mode patterns

---

Built as a Claude Code skill. Runs entirely in your local Claude Code session — no external services, no data leaves your machine.