Use Claude Code's New Auto Mode to Build a…

Every time a job application comes in, someone has to read it, decide if it's worth a call, and either respond or let it sit. That manual triage takes time you don't have when you're hiring fast. This build gives you an automated pipeline: applications arrive, an AI scores each one against your criteria, strong matches ping your Slack channel instantly, and every applicant lands in a Google Sheet — all without you reading a single raw submission first.

How it flows

Get Claude Code running on your machine

Claude Code is Anthropic's AI agent that runs inside your computer's file system — it reads, writes, and organises files for you in response to plain-English instructions. It is not a chat window in a browser; it works in a folder on your computer. You need a paid Claude plan (Pro at $20/month already includes it). Download the desktop app from claude.com/download, install it like any normal app, then open it and sign in with your Claude account. Once inside, click the Code tab to open a coding session.

Claude Code Desktop · opening a session

1 Download the desktop app

Go to claude.com/download and grab the installer for macOS or Windows. Install it like any normal app, then sign in with your Claude account. Claude Code download page →

2 Open a coding session

Code

3 Start your session in the right folder

Working directoryA new empty folder, e.g. ~/recruiting-workflow

Claude reads this folder to understand your project

Create a CLAUDE.md file that tells Claude about your job

CLAUDE.md is a plain text file you put in your project folder. Claude Code reads it automatically every time you start a session — think of it as a permanent sticky note that tells the AI what you are building and what the rules are. Create a new file called `CLAUDE.md` in your folder, then paste in the details of the role you are hiring for: job title, must-have skills, nice-to-haves, and any deal-breakers. This context shapes every scoring decision the AI node will make later.

Your text editor · CLAUDE.md contents

1 Example CLAUDE.md content to paste in

# Recruiting Workflow Role: Senior Product Designer Must-haves: 5+ years UX, Figma, user research Nice-to-haves: SaaS product experience, design systems Deal-breakers: no portfolio link Output language: English

ReturnsClaude loads this on every session start

2 Why this matters

CLAUDE.md stops Claude from giving generic answers. With it, every scoring prompt it writes will already know your specific hiring criteria. Claude Code memory docs →

Ask Claude Code to scaffold the n8n workflow JSON

Now comes the part that the new auto mode makes painless. In the old default mode, Claude Code would stop and ask your permission before writing each file — click approve, click approve, click approve. Auto mode replaces those interruptions with a built-in safety classifier that checks each action itself, so Claude can write your whole workflow file in one go. Just type your request in plain English. Claude will create a valid n8n workflow JSON file — a structured file that n8n can import directly, containing all the nodes and connections pre-wired.

Claude Code · chat input

1 Type this into the Claude Code chat

Create an n8n workflow JSON file called candidate-screener.json. It should: 1) start with a Webhook trigger node, 2) pass the submission to an AI Agent node that scores candidate fit 1-10 using the role criteria in CLAUDE.md and returns JSON {score, reason}, 3) use an IF node to branch if score >= 7, 4) send a Slack message on the true branch, 5) append every applicant to Google Sheets on both branches. Use placeholder credentials.

Returnscandidate-screener.json written to your folder

2 Auto mode in action

Because auto mode is now the default, Claude writes the file without asking you to approve every step. You will see its actions listed in real time, but no clicking 'approve' for each one. Anthropic auto mode announcement →

Import the workflow into n8n

n8n is a visual automation tool — think of it as a flowchart builder where each box (called a node) does one job, and arrows connect them. You do not need to code anything; you just configure each box. Sign up for a free trial at n8n.io/cloud (takes under two minutes, no install needed). Once inside, you will see a blank canvas. Use the Import option to load the file Claude just wrote.

n8n Cloud · main canvas

1 Create a new workflow

+ New workflow

2 Import the file Claude wrote

SFS Start from scratchUAT Use a templateIFF Import from file

3 Select your file

Browse to the candidate-screener.json file Claude created in your project folder and open it. All five nodes appear on the canvas, already connected. n8n import guide →

Connect your credentials (Webhook URL, Slack, Google Sheets)

Each node that talks to an outside service needs a credential — basically a password or API key that proves n8n is allowed to use that service on your behalf. You only set these up once; n8n saves them for future workflows. Click each node on the canvas to open its settings panel, then follow the in-app prompts to connect your accounts. The Webhook node gives you a unique URL you will paste into your application form.

n8n · node settings panels

1 Webhook node — copy your intake URL

HTTP MethodPOST

Test URLShown at top of panel — copy this

Switch to Production URL before going live

2 Connect Google Sheets

Google Sheets

AOU Append or Update RowGoogle Sheets

GR Get Row(s)Google Sheets

3 Map applicant fields to sheet columns

name→Name

score→AI Score

reason→Reason

Ignore blank values

4 Slack node — set your alert channel

Channel#hiring-alerts

MessageNew strong candidate: {{name}} — Score {{score}}/10. {{reason}}

Only fires when IF node returns true

Configure the AI Agent node's scoring prompt

The AI Agent node is the brain of the workflow. It receives the raw form submission and returns a structured score. Click the AI Agent node on the canvas to open it. Inside, you will see a System Prompt field — this is where you tell the AI what role it is playing and what to return. The node also has a Chat Model sub-slot where you connect an AI provider such as OpenAI or Anthropic's own API; click the small plus icon under Chat Model to add one.

n8n · AI Agent node settings

1 Paste this into the System Prompt field

You are a recruiting assistant. Score this applicant 1-10 for fit against this role: Senior Product Designer (must: 5+ yrs UX, Figma, user research; deal-breaker: no portfolio). Return only JSON: {score: number, reason: string (one sentence)}.

Returns{ score: 8, reason: "Strong Figma portfolio, 6 yrs UX, research background." }

2 Add a Chat Model under the AI Agent

OpenAI Chat Model
or Anthropic · connects in one click

Test with a dummy application, then publish

n8n has a built-in test mode so you can try the whole workflow before real applicants see it. Click Test workflow at the top of the canvas — n8n listens for one incoming request on the Test URL. Submit a fake application to that URL (you can use a free tool like Hoppscotch or just fill in a test Typeform/Google Form you pointed at the webhook). Watch the data flow through each node in real time. When everything looks right, hit Publish in the top bar to switch to the Production URL and make it live.

n8n · canvas top bar

1 Start a test run

Test workflow

2 Send a fake application to the Test URL

Paste the Test URL into Hoppscotch (hoppscotch.io) or a form tool, fill in dummy applicant data, and hit Send. Each node lights up green as it passes data along. n8n workflow testing guide →

3 Go live

Publish

What you get

You end up with a live recruiting pipeline: every job application is scored against your criteria automatically, strong matches (score 7+) ping your Slack channel within seconds, and every applicant — strong or weak — is logged to a Google Sheet you can review at any time. You wrote zero code; Claude Code wrote the workflow file and n8n ran it. Rough effort: ~2–3 hours; Claude Pro ($20/mo, includes Claude Code) + n8n Cloud free trial. Google Sheets and Slack are free. OpenAI API costs pennies per scored application..

Use Claude Code's New Auto Mode to Build a Candidate-Screening Workflow in n8n