How to Tailor Your Resume with Claude
Connect Claude to your resumes with one URL, then let it match a job posting, work in the missing keywords, and save a tailored copy. Setup takes two minutes.
The problem with tailoring
Everyone tells you to tailor your resume to each posting. Almost nobody does it, because doing it well takes twenty minutes per application: read the posting, figure out which of your bullets speak to it, notice that the posting says "Kubernetes" three times while your resume says "container orchestration" zero times, rewrite, re-export, rename the file something you'll recognize later.
That legwork is exactly what an AI agent is good at. The judgment calls, whether a claim is true, whether it sounds like you, whether you actually want the job, are yours. This post shows the setup we built for that split.
What connecting Claude actually gets you
Zelume runs an MCP server at https://www.zelume.io/mcp. MCP is the open standard AI assistants use to talk to outside tools, and once Claude is connected it can see and use eleven Zelume tools: list your resumes, fetch one, lint it against real ATS parsing rules, fetch a job posting, score the match, save a tailored copy, and export a PDF, among others.
Two things worth knowing before you decide whether you care:
Claude only sees what you ask it to work with. Every connection signs in through your own Zelume login and gets its own credential, which you can revoke by removing the connector.
And there is no auto-apply here, on purpose. The agent drafts and scores; it never submits anything anywhere. We think the moment a bot clicks "submit" for you is the moment your application stops being yours.
Setup
Claude.ai and Claude Desktop. Settings, then Connectors, then Add custom connector. Paste https://www.zelume.io/mcp. A browser window opens for sign-in (Google, password, or an emailed code, same as the site). That's the whole thing.
Claude Code. One command in your terminal:
claude mcp add --transport http zelume https://www.zelume.io/mcp
Then run /mcp once to authenticate. The tools show up in every session after that.
Cursor or VS Code work too; the agents page has the exact config for each.
A free Zelume account is enough. If you don't have a resume in Zelume yet, import your existing PDF or Word file first; it takes about a minute.
A real tailoring session
Here's roughly how this goes once it's wired up. You paste a posting link and say what you want:
Tailor my Platform Engineering resume to this posting: [link]. Don't invent anything.
Claude fetches the posting, pulls your resume, and runs the match. The first pass usually looks something like "48% keyword match, missing: Kubernetes, Terraform, on-call rotation." Then it does the thing that used to take you twenty minutes: it reads your actual bullets and notices that your last two roles clearly involved Kubernetes and Terraform, you just never wrote the words down. So it proposes edits that surface what's already true.
This is the part to pay attention to. A good agent run rewrites "managed container infrastructure across 12 services" into "ran Kubernetes infrastructure (12 services, Terraform-provisioned)" because both facts were already in your history. A bad run invents an on-call rotation you never had. Zelume's lint flags claims that look manufactured, but the flag is a speed bump, not a guarantee. Read the diff like you'd read a contract someone else drafted in your name.
When you're happy, one more message:
Looks right. Lint it and save it as a tailored copy.
Claude runs the lint (you'll get a score out of 100 and specific line-level warnings, the same engine behind the site), saves the tailored version under a name like "Stripe · Platform Eng" so your original stays untouched, and can hand you the PDF export.
Total time, maybe three minutes. Most of which you spend actually reading the edits, which is the point.
Things people use it for beyond tailoring
Once the connection exists, a few other requests turn out to be useful:
"Lint all my resumes and tell me which one is weakest" is a good Sunday-evening command. So is "compare my two backend resumes and tell me which one to send for this posting." And "what's in my resume that this posting doesn't care about" is the question that finally gets people to cut their high-school job.
If you keep your resume in git as Markdown or LaTeX, that workflow is coming too: the same engine ships as a local CLI with an agent bridge, currently in development. The hosted server is the version you can use today.
The fine print
The tailored copy is saved to your Zelume account like any other resume, so it counts toward nothing; resumes are unlimited on the free plan. AI actions (the tailoring and linting calls) come out of the free plan's daily allowance of 20, which in practice covers several full tailoring sessions a day.
If something breaks, removing and re-adding the connector fixes most of it. And if you'd rather not connect an agent at all, everything above works manually on the site; the agent just does the legwork faster.
Setup details for every client are on the agents page.
Notes on ATS parsing, internship pipelines, and the things we built Zelume to solve.
Action Verbs for Software Engineer Resumes (With Real Bullet Examples)
Built, shipped, debugged, scaled. The verbs that work on an engineering resume, the ones that backfire, and wo…
11 Synonyms for 'Managed' on a Resume (and When Each Fits)
'Managed' appears on almost every resume, often three times. Here are 11 more precise alternatives, sorted by…
What to Say Instead of 'Responsible For' on a Resume
'Responsible for' describes your job posting, not your work. Here are 12 stronger replacements, matched to wha…