Skip to main content
Glama
Johnnnmai

AI Interview Questions MCP

by Johnnnmai

AI Interview Questions MCP

npm license: MIT

An MCP server that brings 200 graded interview questions with model answers into Claude, Cursor, VS Code, or any MCP client: 100 AI Engineer questions (LLM fundamentals, prompting, RAG, agents and MCP, fine-tuning, evaluation, AI system design, production, coding) and 100 Investment Banking questions (accounting, valuation, M&A, LBO, industry and restructuring). Practice, then run a mock interview with your assistant.

  • 200 questions (ids aie-001 to aie-100 and ib-001 to ib-100), each graded Entry / Mid / Senior, with a model answer

  • 50 AI Engineer deep dives with an answer framework, likely follow-ups and common mistakes

  • Runs offline over stdio, or over Streamable HTTP (see Remote HTTP). No API keys, no account, no telemetry

Questions come from the free AI Engineer Interview Questions bank and the Investment Banking question bank by Valenx Press. Want more IB practice? Get the free IB Top 30 questions guide.

Try it

Ask your assistant:

  • "Give me a 5-question Senior mock interview on RAG, one question at a time."

  • "Search the interview questions for KV cache and show me the model answer."

  • "Quiz me on agents, then score my answer to aie-041 from 1 to 5."

  • "Run a 5-question Investment Banking mock interview on valuation."

  • "Walk me through ib-001 and then ask me a follow-up."

Related MCP server: Interview Coach MCP

Tools

Tool

What it does

list_topics

The 14 topics (9 AI Engineer, 5 Investment Banking) with question counts per level

search_questions

Keyword search over questions and answers. Filters: track, topic, level, limit

get_question

One question by id (aie-026, ib-012), number (26) or slug (what-is-rag): model answer, framework, follow-ups, common mistakes. include_answer=false hides the answer for practice

random_question

1 to 10 random questions with answers hidden. Filters: track, topic, level; seed replays the same set

All tools are read-only.

Prompt: mock_interview (args: topic, level, count, track) runs a full round: one question at a time, a 1 to 5 score against the model answer, a follow-up, and a study list at the end.

Resources: aiq://topics and aiq://question/{id} (for example aiq://question/aie-003).

Tracks: ai-engineer, investment-banking.

AI Engineer topics: llm-fundamentals, prompting, rag, agents, fine-tuning, evaluation, system-design, production, coding.

Investment Banking topics: ib-accounting, ib-valuation, ib-ma, ib-lbo, ib-industry.

Install

Requires Node.js 18 or newer.

Claude Code

claude mcp add ai-interview-questions -- npx -y ai-interview-questions-mcp

Or install it as a plugin, which also adds a mock-interview skill:

/plugin marketplace add Johnnnmai/ai-interview-questions-mcp
/plugin install ai-interview-questions@valenx-press

Claude Desktop

Settings > Developer > Edit Config, then add to claude_desktop_config.json:

{
  "mcpServers": {
    "ai-interview-questions": {
      "command": "npx",
      "args": ["-y", "ai-interview-questions-mcp"]
    }
  }
}

Restart Claude Desktop. You can also download ai-interview-questions-mcp.mcpb from the latest release and open it to install with one click.

Cursor

Add to ~/.cursor/mcp.json (or .cursor/mcp.json in a project):

{
  "mcpServers": {
    "ai-interview-questions": {
      "command": "npx",
      "args": ["-y", "ai-interview-questions-mcp"]
    }
  }
}

VS Code

code --add-mcp '{"name":"ai-interview-questions","command":"npx","args":["-y","ai-interview-questions-mcp"]}'

Windows

If npx is not found, use "command": "cmd" and "args": ["/c", "npx", "-y", "ai-interview-questions-mcp"].

From source

git clone https://github.com/Johnnnmai/ai-interview-questions-mcp
cd ai-interview-questions-mcp
npm install && npm run build && npm test

Then point your client at node /path/to/ai-interview-questions-mcp/build/index.js.

Remote HTTP

The same tools are available over Streamable HTTP for clients that only accept remote MCP servers (for example Claude connectors and ChatGPT apps). The HTTP server is stateless and read-only, needs no auth, and stores nothing.

Hosted endpoint: coming soon. Run your own in one command:

npm run start:http        # http://localhost:3000/mcp

Or deploy the repo to Vercel as is: api/mcp.js plus vercel.json serve the endpoint at /mcp.

Free tools that pair with it

Every tool result ends with a short source line and these links, tagged utm_source=mcp so the site can see visits from this server. Nothing else is tracked. See PRIVACY.md.

Development

npm install
npm run build        # compile src/ to build/
npm test             # stdio client calls every tool, prompt and resource, then the HTTP test
npm run start:http   # Streamable HTTP server on PORT (default 3000), endpoint /mcp
npx @modelcontextprotocol/inspector node build/index.js

data/questions.json is generated from the public files of the question bank repo with scripts/build_data.py. data/ib-questions.json holds the 100 Investment Banking questions (scripts/import_ib.py).

The content is educational interview practice, not a record of questions asked by any specific employer. Corrections are welcome via issues.

License

MIT © 2026 Valenx Press

Related MCP Connectors

Related MCP Servers

  • F
    license
    Not graded
    quality
    D
    maintenance
    A Dockerized MCP server that fetches your GitHub repositories, indexes them into a local vector database, and exposes semantic code search tools to LLM clients for tailored technical mock interviews grounded in your actual code.
    -
  • A
    license
    Not graded
    quality
    A
    maintenance
    Enables offline coding interview practice with original problems, local test harnesses, tiered hints, and streak tracking within MCP-compatible agents.
    8 npm
    MIT
  • A
    license
    Not graded
    quality
    A
    maintenance
    Enables AI IDEs to chat with, search, explain, refactor, and audit a local codebase without uploading source code, returning answers backed by file-and-line source citations, confidence scores, and severity labels, plus audit and board-ready strategic reports. Runs locally and standalone as an MCP server compatible with Claude, Cursor, Windsurf, and DeepSeek Harness.
    629 npm
    2
    MIT