ats-resume-writer
Rewrites resumes to beat ATS screening (Workday, Greenhouse, iCIMS, Taleo) against a specific job description, with strict truthfulness guardrails — never invents dates, metrics, titles, or seniority. Pay-what-you-want access codes ($0 works).
README
ATS Resume Writer — MCP Server
Let your AI agent rewrite resumes that beat applicant tracking systems.
When a user asks Claude, ChatGPT, Cursor, or any MCP-capable agent "tailor my resume to this job posting," this server does the rewrite: keyword-matched to the job description, quantified bullets, ATS-safe structure (Workday, Greenhouse, iCIMS, Taleo) — with strict truthfulness guardrails. The engine never invents dates, metrics, job titles, or seniority levels that aren't in the input resume.
Tools
| Tool | What it does |
|---|---|
rewrite_resume |
Takes resume text + a job description, returns the ATS-optimized rewrite as plain text and/or structured JSON. Requires ACCESS_CODE. |
get_access_info |
No auth. Explains how to get an access code (pay-what-you-want, $0 works), what the engine does, and whether the server is configured. |
Setup
Get an access code — pay-what-you-want at Gumroad ($0 works; the code arrives in the delivery email along with the full resume kit).
Hosted (no install) — recommended
The server is live as a remote MCP endpoint — connect any Streamable-HTTP-capable client directly:
https://ats-resume-writer-app.vercel.app/api/mcp
Or via Smithery: mutamiri/ats-resume-writer. On the hosted endpoint, pass your access code as the access_code argument to rewrite_resume (no env config needed).
Claude Desktop / Claude Code
{
"mcpServers": {
"ats-resume-writer": {
"command": "npx",
"args": ["-y", "tsx", "/path/to/ats-resume-writer-mcp/src/server.ts"],
"env": { "ACCESS_CODE": "your-code-here" }
}
}
}
Or after npm run build:
{
"mcpServers": {
"ats-resume-writer": {
"command": "node",
"args": ["/path/to/ats-resume-writer-mcp/dist/server.js"],
"env": { "ACCESS_CODE": "your-code-here" }
}
}
}
Cursor / other MCP clients
Same shape — stdio transport, ACCESS_CODE in env.
Quick start (development)
npm install
npm run typecheck
npm run build
ACCESS_CODE=your-code npm run test:mcp # live smoke test against the hosted API
npm run stdio # run the MCP server on stdio
How it works
agent ──MCP/stdio──> this server ──HTTPS──> hosted rewrite API ──> Claude
(access code gate,
rate limiting,
truthfulness guardrails)
The server is a thin client: the rewrite engine, rate limiting (10/hour), and guardrails live in the hosted API. Resume content is processed for the rewrite and not stored.
Truthfulness guardrails
The engine refuses to:
- Emit any date not present verbatim in the input
- Invent numbers, percentages, or dollar amounts (unambiguous arithmetic on input numbers is allowed, e.g. "5 days → 2 days" may be restated as "60% reduction")
- Add seniority words ("Senior", "Lead", "Principal", "Director") to job titles the candidate didn't hold
- Compute "X+ years of experience" from work-history dates
A truthful resume without a number beats a fabricated one with a fake number.
Companion kit
The same pay-what-you-want purchase includes the full kit: ATS-safe resume template, 7-step writing guide, 54-bullet quantified bullet bank across 9 roles, cover letter framework with worked examples, LinkedIn checklist, and STAR interview prompts.
License
MIT © SokoTech
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。