Connect write.cv to Your AI Agent with MCP
AI agents are moving from chat boxes into real tools — they can now call external services to actually get things done. write.cv exposes one of those services: a public MCP (Model Context Protocol) server. Connect it to an AI agent and the agent can build a CV, check its ATS score, and hand you a one-click link that opens it in the editor. Here's what MCP is, and how to use it.
What is MCP, in plain terms
MCP (Model Context Protocol) is an open standard that lets AI assistants call external tools in a consistent way. Instead of the model guessing or copy-pasting, it can invoke a real function — "score this CV," "generate a share link" — and get a structured result back.
write.cv runs a public MCP server at write.cv/mcp. It needs no signup and no API key, and it's read-only and privacy-preserving: there's no user-data store to expose, because your CV lives in your browser.
What an agent can do with it
Connected, an agent gets a small set of focused tools: fetch the CV JSON schema and authoring rules, validate a CV file for structural problems, score a CV against ATS criteria (0–100 with prioritized fixes), list the templates, and create a share link.
The standout is create_cv_share_link: it turns a CV into a link that opens directly in the write.cv builder, fully editable. So you can say "build my CV from this background and give me a link," and the agent authors it, scores it, improves it, and hands you a URL — the CV is encoded in the link and never sent to a server.
How to connect it
In Claude Code, run: claude mcp add --transport http write-cv https://write.cv/mcp. In Cursor or VS Code, add a server entry pointing to https://write.cv/mcp. In claude.ai or ChatGPT connectors, add the same remote MCP URL. CORS is open, so browser-based clients work too.
Once it's connected, just ask in natural language — "score this CV," "make a CV from my résumé and give me an edit link" — and the agent uses the tools for you. The full setup guide lives at write.cv/AGENTS.md.
Why this matters for job seekers
It closes the loop between an AI assistant and a real, designed, ATS-scored CV. The agent does the heavy lifting — drafting, tailoring to a job, checking the score — and you get a polished CV you can open, tweak, and export, without copy-pasting JSON or fighting formatting.
And because the agent itself is the intelligence, there's nothing to pay for and no quota to burn: write.cv's tools validate, score, and render; your assistant writes.
Prefer to do it yourself? Nothing changes
MCP is an optional power-user feature. The write.cv builder works exactly as before in your browser — type your CV, watch the live ATS score, use the built-in AI assistant, import a PDF, translate it, and export. The MCP server simply opens the same capabilities to AI agents that want to help you from the outside.