How to Schedule Automated Job Applications in Claude Code

The best job-search routine is the one you don’t have to remember. With ResumeAlign’s MCP server and Claude Code’s scheduled agents, you can run the whole loop — scan new postings, tailor your résumé to each, queue the applications — on a cron, every morning, without opening anything. The tailoring runs on your own model, so it’s hands-free and $0.

Why schedule it

Fresh postings get the most attention in their first day or two. A scheduled agent applies to new matches while they’re new — consistently, without you carving out time. And because the AI runs on your Claude Code subscription, doing it daily costs no more than doing it once.

First, connect (one time)

If you haven’t already, connect Claude Code to ResumeAlign:

claude mcp add --transport http resumealign \
  https://alignresume.com/mcp \
  --header "Authorization: Bearer <your-token>"

The loop the agent runs

Every run, the agent works through this until the queue is dry:

• run_autopilot(keywords, offload=true) — scan the job board for your keywords and queue the matches (no server tailoring).
• next_jobs_to_tailor(10) — pull a batch of new postings (with the cleaned job text + your base résumé).
• tailor each on its own model, then submit_tailored_resume(..., structure) — stores it ready to apply, rendered at $0.
• apply_queued() — submit that batch; repeat.

The scheduled-agent prompt

In Claude Code, create a scheduled agent (routine) that runs on your cadence — say weekday mornings — with a prompt like this:

Every run, run my ResumeAlign job search via MCP, offloaded to you:

1. run_autopilot(keywords="senior data engineer", offload=true)
2. Loop: next_jobs_to_tailor(limit=10). For each job, tailor my
   résumé to it truthfully (mirror the posting's real keywords,
   quantify impact, never invent experience), build the structure,
   and submit_tailored_resume(job_id, text, structure).
3. apply_queued() after each batch.
4. Stop when next_jobs_to_tailor returns no jobs.
Then summarize: how many tailored and applied, and any strong
matches I should follow up on personally.

Now turn that prompt into a recurring routine.

Create the scheduled agent in Claude Code

Claude Code calls these routines — scheduled agents that run on Anthropic’s cloud on a cron. Create one two ways:

From the CLI, just describe it — the /schedule command walks you through name, prompt, schedule, and connectors:

/schedule weekdays at 8am — run my ResumeAlign job search via MCP, offloaded to you

From the web at claude.ai/code/routines → New routine:

• Instructions — paste the prompt above.
• Trigger → Schedule — pick Weekdays, or a custom cron like 0 8 * * 1-5 (8am Mon–Fri). Minimum interval is one hour.
• Connectors — add the ResumeAlign MCP server (https://alignresume.com/mcp) with your Authorization: Bearer <token> header so the routine can call the tools.
• Click Create.

The CLI and web write to the same account, so a routine created in one shows up in the other.

List, edit, delete, run now

• List: /schedule list, or open claude.ai/code/routines.
• Edit the prompt or schedule: /schedule update (pick the routine), or the Edit (pencil) icon on the web.
• Run now (don’t wait for the next slot — great for testing): /schedule run, or Run now on the routine’s page.
• Delete: open the routine on the web and click the trash icon (deletion is done from the web UI).

What needs to be running

Two things make it actually fire:

• Plan: cloud routines need a Claude Pro or Max plan with Claude Code on the web. They run on Anthropic’s infrastructure, so your machine doesn’t need to be on for the routine itself.
• Your browser, for scan + submit: scanning the job board and clicking Apply are done by the ResumeAlign browser extension — the routine triggers them over MCP but can’t reach your browser itself. So keep alignresume.com open with the extension connected and signed in to your job board (e.g. Dice) while the routine runs. The tailoring (the expensive part) is fully autonomous on your model regardless; if your browser isn’t up, the run still tailors and queues everything, then applies the next time the extension is connected.

Keep an eye on it

Every routine run is a session you can open from claude.ai/code/routines to see exactly what it did. And any time, ask your agent: “Show my recent auto-pilot runs and what I spent.” — it calls list_autopilot_runs and apply_stats so you can see volume per run, the keywords searched, and the cost (which, on the offload path, is $0).

Cost

Because every tailoring call runs on your Claude Code subscription, a daily routine adds no per-use AI cost. The free plan covers 30 applications/month; the MCP Plan ($19.99/mo) makes the scheduled loop unlimited.

Frequently asked questions

Can Claude Code run a task on a schedule?
Yes. Claude Code supports scheduled agents (routines) that run a prompt on a recurring cadence. Point one at ResumeAlign’s MCP server and it can scan, tailor, and queue applications every morning without you lifting a finger.

What does the scheduled agent actually do each run?
It calls run_autopilot to scan job boards for your keywords, pulls the new postings with next_jobs_to_tailor, tailors your résumé to each one on its own model, submits them with submit_tailored_resume, and applies the batch with apply_queued — looping until the queue is empty.

What do I need for the scheduled run to actually apply?
Tailoring is fully automated on your model. The final submit step uses the ResumeAlign browser extension, so keep a browser open with the extension connected and signed in to your job board (e.g. Dice) when the schedule fires. Everything up to the submit is hands-free and $0.