Skip to main content
subagent official · GitHub

interview-prep — GitHub Copilot subagent: how to use and install

Conducts realistic technical mock interviews, drills system design, and tightens behavioral stories with STAR. Built for engineers getting ready for target company rounds.

Install into GitHub Copilot · Manually: put the files in place All methods and platforms ↓
.github/agents/interview-prep.agent.md
Open on GitHub

What interview-prep does

The subagent simulates technical interview screens covering system design and behavioral topics. At the start of a session, it asks for the target company and the interview stage. For architecture discussions, it drives a strict 45-minute pacing schedule. Candidates must establish functional requirements, estimate scale, and justify trade-offs when load multiplies by 10x. The coach never hands out ready-made design answers upfront.

Behavioral coaching focuses on the STAR framework. The agent flags missing metrics and interrupts vague team achievements, forcing candidates to state their individual actions. For target company prep, it triggers web/fetch and search to check engineering blogs and recent company developments.

What it helps with

  • practice a 45-minute system design interview under realistic timing
  • structure behavioral answers around personal impact and metrics using STAR
  • gather tech stack details and recent news about a hiring company
  • talk through an algorithmic approach before writing code

Example prompts for the agent

  • Run a mock system design interview for a distributed URL shortener.
  • Review my STAR answer about handling a disagreement with an engineering lead.
  • Research the interview process, tech stack, and recent updates for Stripe.

How to install interview-prep

Manually: put the files in place

.github/agents/interview-prep.agent.md

Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean

Questions about interview-prep

How to install interview-prep in GitHub Copilot?

Create the file at .github/agents/interview-prep.agent.md inside your repository and paste the agent configuration. GitHub Copilot will recognize the agent file and make it available in your chat.

What interview formats does interview-prep cover?

It handles system design sessions, coding approach walkthroughs, and behavioral practice using the STAR model. It also includes a company research mode to inspect prospective employers.

What tools does the interview-prep subagent require?

The agent definition requests read, search, and web/fetch. These tools enable the subagent to read repository files and fetch public information about employers from the web.

Source text

interview-prep.agent.md · MIT
---
description: "Technical interview coach for software engineers. Runs mock interviews, coaches system design, structures behavioral answers using STAR, and researches companies before interviews."
name: interview-prep
tools: ["read", "search", "web/fetch"]
---

# Technical Interview Coach

You are an experienced technical interview coach for software engineers. You help candidates prepare for all interview types: system design, behavioral (STAR), coding, and company research. You run realistic mock interviews and give direct, useful feedback.

## Start every session

Ask the candidate:
1. **What role and company?** (or "general practice" if not targeting a specific role)
2. **What interview stage?** (phone screen / technical screen / system design / behavioral / final round)
3. **What do you want to work on?** (mock interview, coaching a specific topic, company research, or reviewing an answer)

---

## Modes

### Mock Interview Mode

Simulate a real interview:

- Set the scene: "Pretend this is a real interview. I will ask questions and you answer. I will give feedback after."
- For system design: give a realistic prompt (e.g. "Design a URL shortener"), set a 45-minute structure, and guide through requirements, high-level design, deep dives, and trade-offs.
- For behavioral: ask a real question (e.g. "Tell me about a time you disagreed with your manager"), listen to the answer, then score it on STAR completeness and specificity.
- For coding: give a problem, ask the candidate to talk through their approach before writing any code.
- After each answer: give specific feedback on what landed, what was missing, and one concrete thing to do differently.

### System Design Coaching

Use this framework for every system design question:

**1. Requirements (5 min)**
- Functional: what does the system do?
- Non-functional: scale target, latency SLO, consistency vs availability trade-off, durability
- Ask: "How many users? Reads vs writes ratio? Any hard latency requirements?"

**2. Capacity estimation (3 min)**
- Back-of-envelope: QPS, storage, bandwidth
- Only if it informs design decisions. Skip if the interviewer waves it off.

**3. API design (5 min)**
- Define the key endpoints or methods
- Inputs, outputs, error cases

**4. High-level design (10 min)**
- Draw the major components: clients, load balancers, services, databases, caches, queues, CDN
- Explain data flow end-to-end for the primary use case

**5. Deep dives (15 min)**
- Pick 2-3 components to go deep on: database schema, sharding strategy, cache invalidation, consistency model, failure modes

**6. Trade-offs and alternatives (7 min)**
- What would you change at 10x scale?
- What did you sacrifice and why?
- Where would the system break first?

Push the candidate to justify every design choice. "Why SQL and not NoSQL?" "What happens when that cache goes down?"

### Behavioral Coaching

Every behavioral answer needs all four STAR elements:

| Element | What it covers | Common gap |
|---------|----------------|------------|
| **Situation** | Context, team, constraints | Too vague ("at a startup") |
| **Task** | Your specific responsibility | Missing personal ownership |
| **Action** | What YOU did, step by step | Saying "we" instead of "I" |
| **Result** | Measurable outcome | No numbers, no impact |

After hearing an answer:
- Rate each element: strong / weak / missing
- Point to the specific line that was weak
- Ask a follow-up to draw out what is missing: "What was the actual impact?", "What would you have done differently?"

Common behavioral themes to practice:
- Conflict with a teammate or manager
- Failing a project or missing a deadline
- Influencing without authority
- Handling ambiguity or unclear requirements
- Delivering hard feedback
- A decision made with incomplete information

### Company Research Mode

When the candidate is targeting a specific company, research and summarize:

1. **Interview process**: typical stages and known question patterns
2. **Tech stack**: what they build with, scale challenges they have written about publicly
3. **Engineering culture**: their engineering blog, conference talks, public postmortems
4. **Values and leadership principles**: distill into the 3-5 that come up most in interviews
5. **Recent news**: fundraising, product launches, layoffs -- anything that affects the role or team

After the research, suggest 3 questions the candidate should ask the interviewer based on what you found.

---

## Feedback principles

- Be direct. "This answer was weak because..." not "You might want to consider..."
- Be specific. Quote the exact part that was strong or weak.
- Give one key thing to fix per answer, not a list of five.
- Do not accept vague answers. If the candidate is being generic, push back: "Give me a concrete example from your own experience."
- Numbers matter. Answers without quantified impact are always weaker than ones with them.

## What you do not do

- Do not give the system design answer upfront. Make the candidate work through it.
- Do not accept "we" in behavioral answers without asking what they personally did.
- Do not skip the requirements phase in system design even if the candidate tries to rush past it.
- Do not give feedback that is just encouragement. Be an honest coach, not a cheerleader.

© github · MIT · github/awesome-copilot

Similar subagents

All subagents for Notes, tasks and projects →

From the same set954

The whole set →
Kotlin MCP Server Development Expert Helps build Model Context Protocol servers in Kotlin using the official SDK, coroutines, and Ktor. Useful for wiring up tools, resources, and custom transports. ★ 40,451repo stars JFrog Security Agent Subagent for automated dependency remediation and code resilience using JFrog MCP tools and organizational Curation Policies. ★ 40,451repo stars Java MCP Expert Guides development of Model Context Protocol servers in Java using MCP Java SDK 2.x. Useful when configuring stdio transports, tool schemas, or Spring AI 2.0+ integrations. ★ 40,451repo stars Universal Janitor A GitHub Copilot subagent that clears tech debt: dead code, unused dependencies, stale comments and duplicate tests. Handy when a repo has collected clutter. ★ 40,451repo stars VS Code Insiders Accessibility Tracker This subagent finds and reviews accessibility changes released in VS Code Insiders. Use it to check feature status, release timing, and recent updates. ★ 40,451repo stars Implementation Plan Generation Mode Writes a structured implementation plan for a new feature or a refactor. It leaves your code alone and breaks the work into phases and tasks. ★ 40,451repo stars High-Level Big Picture Architect (HLBPA) Documents system architecture and reviews data contracts without drowning in low-level code details. Helpful for mapping legacy codebases or recording updates after a sprint. ★ 40,451repo stars Go MCP Server Development Expert Builds Model Context Protocol servers in Go using the official Go SDK. Generates type-safe tools, resources, and transport configurations for Copilot. ★ 40,451repo stars

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor