Job applications are repetitive, high-volume work: read a posting, re-tailor your résumé to it, format it, submit. It’s exactly the kind of loop an AI agent should run for you. ResumeAlign now exposes its whole pipeline to your own AI agent over the Model Context Protocol (MCP) — so Claude Code can score jobs, tailor your résumé, and auto-apply, with the AI running on your Claude subscription instead of a metered API.
You’ve probably seen BYOK — Bring Your Own Key — where a tool runs on your API key and you pay the provider per token. BYOM goes one step further: you bring your whole model. When you drive ResumeAlign from Claude Code, the résumé writing happens inside your Claude Code session — which you already pay for at a flat rate — so there’s no per-use AI charge at all. ResumeAlign’s only remaining work is scraping the posting, rendering the formatted PDF, and submitting; none of that touches an LLM.
1. Mint an access token. Log in at alignresume.com and create a token (the same one the browser extension uses). It scopes the connection to your account.
2. Add the MCP server. In your terminal:
claude mcp add --transport http resumealign \
https://alignresume.com/mcp \
--header "Authorization: Bearer <your-token>"That’s it. The anonymous tools and the prompts work with no token; your saved résumés and the apply queue use the token above.
The server exposes the parts only ResumeAlign can do as tools, and pushes the expensive writing to your model as prompts:
analyze_job_match — scrape a posting and score your résumé against it (keywords, seniority, gaps).tailor_resume / strengthen_resume (prompts) — your model rewrites the résumé to the job, truthfully.format_resume_pdf / submit_tailored_resume — ResumeAlign renders a clean, ATS-friendly PDF (you can hand it the structure for a pixel-perfect, zero-LLM render).run_autopilot, next_jobs_to_tailor, apply_queued — scan a job board, pull untailored postings, and submit applications.list_autopilot_runs, apply_stats — see how many you’ve applied to and what it cost.You can just talk to it. For example:
You: Score my résumé against this job: <url>
Claude: 84% match — strong on Kafka/streaming, missing
"Flink" and an explicit FastAPI mention.
You: Tailor it to that posting and format it as a PDF.
Claude: [rewrites on its own model] → calls format_resume_pdf
→ here’s your tailored, ATS-clean PDF.Or hand the whole job board to it: “Do an auto-pilot for ‘mcp developer’, remote only.” The agent scans the board, tailors each match on its model, and queues them to apply.
On the offload path the tailoring is done by your Claude Code, so each application costs ResumeAlign nothing in AI and costs you nothing in API credits — only your flat Claude subscription. A server-driven run (the classic way) tailors on our model and is billed; the agent-driven run is free. Same résumés, same auto-apply — the difference is whose model does the thinking.
Next: put it on a schedule so your job search runs itself every morning.
What is BYOM (Bring Your Own Model)?
BYOM means the AI reasoning runs on your own model — your Claude Code subscription — instead of on the vendor’s servers or your metered API key. With ResumeAlign’s MCP server, your Claude Code agent does the résumé tailoring, so there is no per-token Anthropic API cost at all. It’s the step beyond BYOK (Bring Your Own Key), where you still pay Anthropic per token.
How do I connect Claude Code to ResumeAlign?
Mint an access token (log in at alignresume.com and POST /api/auth/extension-token), then run: claude mcp add --transport http resumealign https://alignresume.com/mcp --header "Authorization: Bearer <token>". Anonymous tools and prompts work without a token; account features (your saved résumés, the apply queue) need it.
Does auto-apply cost anything?
Through the MCP offload path the tailoring runs on your Claude Code subscription, so ResumeAlign spends $0 on AI and you spend $0 in API credits — only your flat Claude subscription. The free plan includes 30 MCP applications per month; the MCP Plan ($19.99/mo) makes it unlimited.