360 Secure SkillHub
360 安全技能中心

ATSPrep: Interview prep

调用 ATSPrep 工具,依据职位描述和简历生成可能的面试问题,逐题提问并等待用户回答,再按 STAR 结构点评、指出缺失要点并给出更强版本答案;支持 MCP 工具、内置 Node 脚本或直接 HTTPS 调用。

分类:知识
devabubakar
devabubakar
热度 12.26
最近更新 2026-11-04 02:10:42

安装

发给 Agent
帮我安装这个技能:npx secure-skills-cli add devabubakar/atsprep-interview-prep
兼容性
通用兼容
联网要求
需要联网
数据出网
必须出网
Metadata
name
interview-prep
description
Run a mock job interview with ATSPrep's question bank - pick likely questions from the job post and resume, ask them one at a time, and give STAR-based feedback with a stronger version of each answer. Use when someone has an interview coming up, wants likely interview questions, wants to practise answers, or asks to improve a specific answer.
license
MIT
compatibility
Needs network access to https://atsprep.com. The bundled script needs Node 18+.
metadata
{"author":"ATSPrep","homepage":"https://atsprep.com/mcp","version":"1.0.0"}

Interview prep

A mock interview, one question at a time. ATSPrep's web version, with spoken practice, is at https://atsprep.com/tools/interview-questions.

How to call ATSPrep

Use whichever is available, in this order:

  1. ATSPrep MCP tools. If the ATSPrep MCP server (https://atsprep.com/mcp) is connected, call its tools by name. Some clients add a prefix, e.g. mcp__atsprep__check_resume.
  2. The bundled script (needs Node 18+ and network access):
    node scripts/atsprep.mjs <tool> key=value key=@path/to/file.txt
    
    key=@file sends a file's text, which is the safe way to pass a resume or job post. Write the user's text to a file in the current working directory first, and delete it when you're done. node scripts/atsprep.mjs list lists the tools.
  3. Plain HTTPS (no Node): POST JSON-RPC to https://atsprep.com/mcp with headers Content-Type: application/json and Accept: application/json, text/event-stream, and the body {"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"<tool>","arguments":{...}}}.

Arguments and limits for every tool are in references/tools.md. What leaves the machine is in references/privacy.md. Tell the user before sending their text to an AI tool the first time.

Steps

  1. Ask for the job title and, ideally, the job post (for a link, use fetch_job_post) and their resume.
  2. Call interview_questions with what you have (job_description, resume_text) and limit around 8. With neither, it returns common questions; tailor them to the job title yourself. Use kind (posting, resume, about, behavioural, role or closing) to focus on one group if the user asks.
  3. Share the STAR tips the tool returns in a few lines, then ask one question and wait for the answer. Don't list every question unless asked.
  4. Call review_interview_answer (AI) with question, answer and kind: story for "tell me about a time", pitch for "tell me about yourself", plan for "how would you...", number for salary, close for "any questions for us?". Share what worked, what was missing (situation, task, action, result) and the stronger version. If the AI part was skipped, the rule-based feedback still applies.
  5. Let them retry or move on. At the end, sum up their two or three biggest patterns.

Coaching rules

  • The stronger answer may only use facts from the user's own answer. If a result needs a number, ask what it was.
  • Keep feedback short and specific. One fix per answer beats ten.
  • Never tell the user to claim experience they don't have.