AgentSkillsCN

get-a-job

一款端到端的求职助手，可帮助用户查找、追踪并申请各类职位。当用户希望：(1) 根据自己的简历、职业偏好、目标岗位、行业方向及工作地点等条件设置求职计划；(2) 在网络上搜索符合自身要求的新职位信息；(3) 维护一份正在投递的职位清单，同时另建一张心仪公司的名单；(4) 针对特定公司量身定制申请材料（简历、求职信、跟进邮件）；(5) 获取每日或每周的求职进展总结，并可选择以LinkedIn或X平台的帖子形式进行呈现时，即可使用此技能。触发词包括：“找份工作”、“求职”、“帮我找找工作”、“开始扫描”、“我们去投[公司]吧”、“更新我的心仪公司”、“每日总结”、“每周回顾”、“设置我的求职计划”，或任何有关查找空缺岗位、追踪申请进度、准备申请材料的请求。

SKILL.md
--- frontmatter
name: get-a-job
description: >
  End-to-end job search assistant that helps users find, track, and apply to jobs.
  Use this skill when the user wants to: (1) set up a job search with their resume,
  preferences, target roles, industries, and location requirements, (2) scan the web
  for new job listings matching their criteria, (3) maintain a spreadsheet of active
  listings and a separate spreadsheet of dream companies they admire, (4) prepare
  tailored application materials (resume, cover letter, follow-up email) for specific
  companies, (5) get daily or weekly summaries of their job search progress, optionally
  formatted as LinkedIn or X posts. Triggers include phrases like "get a job", "job search",
  "find me jobs", "run a scan", "let's apply to [company]", "update my dream companies",
  "daily summary", "weekly recap", "set up my job search", or any request to find open
  roles, track applications, or prepare application materials.

Get a Job

A structured job search workflow that organizes everything in a Get a Job folder on the user's desktop.

Workspace Structure

code
Get a Job/
├── Master/
│   ├── resume.md          (master resume — the source of truth)
│   └── portfolio_links.md (portfolio URLs, GitHub, site, etc.)
├── search_config.csv      (roles, industries, locations, preferences)
├── Job_Listings.csv       (active openings found via web search)
├── Dream_Companies.csv    (aspirational companies with news + careers info)
├── Applications/
│   └── [CompanyName]/     (one folder per company applied to)
│       ├── resume_tailored.md
│       ├── cover_letter.md
│       └── followup_email.md
└── Summaries/
    ├── daily_YYYY-MM-DD.md
    └── weekly_YYYY-MM-DD.md

Workflows

1. Setup — "Set up my job search"

Run on first use or when user wants to reconfigure.

  1. •Ask for resume (file upload or paste). Save to Master/resume.md.
  2. •Ask for portfolio links or references. Save to Master/portfolio_links.md.
  3. •Ask for target roles, industries, and keywords.
  4. •Ask for location preferences:
    • •In-person locations (city, zip, country)
    • •Remote OK? Hybrid OK?
    • •Exceptions (e.g., "in-person if within 20 min of [zip], remote otherwise")
  5. •Ask for optional salary minimum.
  6. •Run scripts/init_workspace.py <path> to create the folder structure.
  7. •Write preferences to search_config.csv.

Keep questions conversational — ask the most important ones first, follow up as needed. Do not overwhelm with all questions at once.

2. Scan — "Run a scan" / "Find me jobs"

Batch job search the user triggers whenever they want.

  1. •Read search_config.csv for current preferences.
  2. •Read Job_Listings.csv to know what's already tracked.
  3. •Use web search to find new listings. See references/search-strategies.md for query patterns and source priority.
  4. •For each result, gather: company, role, type, location, URL, contact info, salary range.
  5. •Rate match quality 1-5 based on how well it fits user preferences.
  6. •Use scripts/add_listing.py to append new listings (handles deduplication).
  7. •Present a summary of new findings to the user.
  8. •Suggest any dream company additions based on interesting companies encountered.

Run multiple varied search queries (different role titles, industries, locations) to ensure broad coverage. Aim for 5-10 searches per scan session.

3. Dream Companies — "Update my dream companies"

Maintain a list of companies the user admires, even if they have no open listings.

  1. •Ask the user to name companies they like, use, or admire. Suggest others in their industries.
  2. •For each company, use web search to find:
    • •Careers page URL
    • •Jobs or hiring email
    • •Recent news (past 3 months) — summarize in 2-3 sentences
    • •Company direction and trajectory — summarize briefly
    • •Any known contacts
  3. •Use scripts/update_dream_company.py to add or update entries.
  4. •Periodically (during scan sessions), refresh recent_news and direction for existing dream companies.

See references/csv-schemas.md for column definitions.

4. Apply — "Let's apply to [company]"

When the user is ready to apply to a specific company.

  1. •Read the master resume from Master/resume.md.
  2. •Read the job listing details from Job_Listings.csv or Dream_Companies.csv.
  3. •If needed, web search for more details about the role and company.
  4. •Create Applications/[CompanyName]/ folder (underscores for spaces).
  5. •Generate three files per references/application-materials.md:
    • •resume_tailored.md — master resume reordered and tuned for this role
    • •cover_letter.md — professional, specific, under 400 words
    • •followup_email.md — for sending 5-7 days post-application
  6. •Present the files and walk the user through suggested changes.
  7. •Update the listing's status to applied in the relevant CSV.

5. Summaries — "Daily summary" / "Weekly recap"

Generate progress reports, optionally formatted for social sharing.

  1. •Read Job_Listings.csv and Dream_Companies.csv for current state.
  2. •Compare against previous summaries in Summaries/ to identify what's new.
  3. •Generate summary per references/summary-templates.md.
  4. •If user requests it, append LinkedIn and X post drafts to the summary.
  5. •Save to Summaries/daily_YYYY-MM-DD.md or Summaries/weekly_YYYY-MM-DD.md.

Tone: professional, forward-looking, momentum-focused. Never negative or desperate.

Key Guidelines

  • •All output files are .md or .csv — no .docx unless explicitly requested.
  • •Always read the master resume before tailoring; never invent experience.
  • •Keep cover letters under 400 words, follow-up emails under 100 words.
  • •When searching, use concise 1-6 word queries for best results.
  • •Deduplicate listings before adding — the scripts handle this automatically.
  • •Leave contact fields blank rather than guessing.
  • •Refresh dream company news during scan sessions to keep info current.
  • •Respect the user's location preferences including conditional rules.