Connect Claude Desktop to Your Resume in 2 Minutes (MCP Setup Guide)

Imagine asking Claude: "Why didn't I match the senior backend role I analyzed last week?" — and getting an answer grounded in your actual resume and that exact vacancy, not a textbook list of "common reasons candidates get rejected."

That's what becomes possible when you connect ResumeQuick to Claude Desktop, Claude Code, or Codex via the Model Context Protocol (MCP). Two minutes of setup, zero coding, and Claude becomes a career advisor that actually knows you.

Why generic AI gives generic advice

Drop "what should I improve in my resume?" into a fresh Claude or ChatGPT chat and you'll get a perfectly reasonable, perfectly forgettable list: add metrics, tailor to the job, use action verbs. All true. None of it specific to your resume, your industry, or your recent vacancy applications.

The reason is mundane: the model has never seen your data. It has read millions of resumes during training, but it has no idea what your resume says, where you applied yesterday, or how you scored on your last mock interview.

MCP fixes that. It's a standard, opened by Anthropic and adopted across the AI tooling ecosystem in 2025–2026, that lets an AI client (Claude Desktop, Claude Code, Codex, Cursor, and others) call tools exposed by a server. Our MCP server at mcp.resumequick.app exposes a focused toolset that lets Claude both read and update the slice of your data that matters for job search — and nothing else.

What Claude can do with your data

Claude connects to a focused toolset — some tools read your data, others make the changes you ask for. The read tools, one job each:

• get_profile_snapshot — Compact profile view: name, summary, top skills, last two roles, location
• list_resumes — List of your saved resumes with titles and last-updated timestamps
• get_resume — Full text of a single resume (you reference by id)
• get_recent_vacancies — Last N vacancy analyses with match percentage and verdict
• get_interview_results — Last N completed mock interviews with score and key feedback
• get_recent_activity — Chronological feed of profile changes in the last N days

Claude calls these when it decides they're useful, or when you ask it to — "use my ResumeQuick data to answer this." On top of reading, Claude can also make changes on your behalf: edit a resume section, tailor a resume to a specific vacancy, create a new resume version, rename or switch your active resume, adjust styling, and update your career notes. What it can't do: send messages or emails, or touch your password and account settings. When you connect Claude Desktop or Claude Code, the client shows each action and asks you to approve it before it runs — so nothing changes without you seeing it.

Six prompts that show why this matters

Once connected, here are six things you can ask Claude that are impossible (or just generic) without your real data:

"Roast my latest resume — what would a senior recruiter cut, and why?"

Claude pulls the document, reads it end to end, and gives you critique grounded in your phrasing and your experience density. No template advice.

"Why didn't I match the platform engineer vacancy I analyzed yesterday?"

Claude reads the analysis, sees the gap, and explains in plain English which of your experiences read weakly against which job requirements.

"Quiz me on the topics where I scored lowest in my last mock interview."

Claude finds the weakest dimensions in your interview results and runs you through follow-up questions in those areas. Spaced repetition, but for behavioral and technical interview prep.

"What changed in my profile in the last 30 days, and what's a fair self-assessment of progress?"

Claude pulls the activity feed and gives you a Monday-morning recap of edits, vacancies analyzed, interviews completed.

"Look at my last three vacancy analyses. Is there a pattern in the kinds of roles I'm matching well or badly?"

Pattern recognition across your real applications, not theoretical career-planning advice.

"Based on my profile, what 3 skills would unlock the next tier of roles?"

Claude grounds the recommendation in what you actually have and what the vacancies you've been analyzing actually demand.

The pattern is the same: Claude does what Claude is good at — reasoning, framing, and clear writing — but on top of your facts.

Setup in 3 steps

1. Create an API key in your ResumeQuick profile

Sign in at resumequick.app, open Profile → API Keys, click Create key, and give it a name you'll remember (e.g. "Claude Desktop on laptop"). You'll see the plaintext key once — it starts with rq_live_…. Copy it now and keep it safe; we don't store the plaintext.

The page also shows a pre-filled config snippet for each supported client right next to the key — so you can usually skip step 2 and copy from there directly. If you're following along here, read on.

2. Paste the snippet into your client

Claude Desktop — open ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows) and add:

{
  "mcpServers": {
    "resumequick": {
      "type": "http",
      "url": "https://mcp.resumequick.app/mcp",
      "headers": {
        "Authorization": "Bearer rq_live_..."
      }
    }
  }
}

Replace rq_live_... with the key you just copied. Save the file and restart Claude Desktop.

Claude Code (CLI) — one command:

claude mcp add --transport http resumequick https://mcp.resumequick.app/mcp \
  --header "Authorization: Bearer rq_live_..."

Codex / Cursor / other HTTP-streamable MCP clients — same JSON shape as Claude Desktop, dropped into your client's MCP config file.

3. Ask away

Open a fresh chat. Try:

"Use my ResumeQuick profile. Look at the last vacancy I analyzed, and tell me the three biggest gaps between my resume and that role."

Claude will call get_recent_vacancies, then get_resume, then write the answer. You'll see the tool calls happening in the UI (Claude Desktop shows them by default; in Claude Code they're visible in verbose mode). The first time it works, it feels a little magical.

Privacy: what we promise

Three commitments worth being explicit about:

1. The key is yours alone. It's tied to your account, gives access only to your own data, and is revocable from Profile → API Keys at any time. Revocation takes effect within seconds — if a laptop walks off, you can kill the key from your phone.
2. No mining. ResumeQuick doesn't analyze the prompts Claude sends or the responses Claude generates. Our server logs the tool calls (which tool, when) for abuse prevention and observability, but not the conversation around them — that lives entirely between you and Anthropic / OpenAI.
3. Your data, and only yours. A key can only ever touch your own ResumeQuick data — never another user's, and never your password or account settings. Every change Claude makes runs through your client, which shows the action and asks you to approve it first — so anything destructive, like deleting a resume, is yours to confirm.

The plaintext key is shown exactly once on creation — we store only a hash. If you lose it, revoke and create a new one.

What's next

What ships today is the floor, not the ceiling. Three threads we're pulling on:

• Scoped, read-only keys. Every key today has full access to your own data. A read-only key — for handing your profile to a tool or coach you don't want making changes — is what we're building next.
• More clients. Anything that speaks the standard MCP HTTP streamable transport works today. As clients add MCP support (Cursor's MCP integration, Gemini, others), our server is already compatible — no work required on your side.
• B2B for career coaches. A scoped variant where a coach can connect to a candidate's profile (with explicit consent) and run the same tooling on the candidate's behalf. If you're a coach who'd use this, tell us.

Try it

If you've read this far, the only thing standing between you and a Claude that knows your career is two minutes of config.

Get your API key →

Or jump straight to the step-by-step setup guide with copy-pasteable snippets for every supported client.