Is this safe for my LinkedIn account?

Yes — LinkedIn automation runs supervised only, meaning you are present and watching every time. The agent stops instantly on any security challenge or CAPTCHA and never runs LinkedIn unattended, which is the behavior that gets accounts banned. It also never types your password or creates an account; it uses your own already-logged-in Chrome session via the Claude in Chrome extension.

Does it fabricate or exaggerate anything on my CV?

No. Your CV lives as a Python dict of your real experience, and the cv-tailor agent may only re-emphasize or reorder what's already in that dict per job description — it cannot invent skills or experience. You also give it a no-claim list of technologies you don't actually know, which it's forbidden from ever mentioning. Every generated PDF is rendered to images and visually QA-checked before use.

Do I need an API key to use this?

Only one optional key: an Apify API key for zero-token SEEK and job-board scraping, and the free tier is enough for most usage. Everything else runs on your existing Claude Code login — there's no separate LLM API key to manage, and the agent never reads any credential besides the optional Apify key in your local .env file.

What does a run cost?

A typical full run — search, filter, tailor a CV, fill and submit a form — costs a few cents of API usage per applied job. The pipeline is deliberately split so Opus (the more expensive model) only runs the two steps that need judgment — filtering and CV tailoring — while Sonnet handles the fast, mechanical steps like searching and form-filling, and plain Python scripts do zero-token fetching wherever possible.

Does it work on Windows and Linux, or just macOS?

The dashboard is a dependency-free Node.js server and the scripts are plain Python, so both run on any OS Node 18+ and Python 3.9+ support, including Windows and Linux. The one platform-specific prerequisite is poppler for CV visual QA — macOS installs it with brew install poppler, Linux with apt install poppler-utils.

What happens when a job application has a CAPTCHA?

The agent never solves or bypasses CAPTCHAs or bot-checks — that's a hard, non-negotiable rule. When it hits one, or any Cloudflare-fronted or account-required application (like Workday), it prepares everything it can and hands the job back to you on the dashboard for a manual submit, rather than trying to get around it.

What exactly do I get when I buy this?

One template document containing every project file — 5 sub-agent definitions, 3 slash commands, 4 scripts, the dashboard, and the rulebook — ready to paste into an empty Claude Code folder, plus the full numbered setup guide unlocked on this page and lifetime updates as the template improves. There's nothing else to install or subscribe to beyond the prerequisites listed above.

Do I need a paid Claude plan?

A Claude Code Pro or Max plan is recommended so you have enough usage headroom for daily runs, but the agent works on any plan that can run Claude Code and the Claude in Chrome extension. Actual per-run cost is small — typically a few cents of API usage per applied job — since only the judgment steps (filtering, CV tailoring) run on the more expensive model.

Job Automation

Apply Jobs Automatically — AI Agent

5 AI agents in Claude Code that search jobs, filter them by your rules, tailor your CV per job description, and fill + submit applications in your own browser — while you watch from a live dashboard.

Claude Code
AI Agents
Job Search Automation
Auto Apply
Resume Tailoring
Cover Letter
Browser Automation
ATS
LinkedIn
SEEK
Indeed
Job Hunting

An autonomous job-application pipeline that runs inside Claude Code. It searches job platforms, filters roles against your criteria, tailors a CV per job description, fills out application forms in your own browser, and submits them — while you watch from a live local dashboard.

Open a dashboard, pick a role and a platform, click Start Applying — the agent does the rest, live, in your own Chrome.

What you get

Dashboard (localhost:4747)
   │  Role ▾  Platform ▾  Location  Work-mode ▾   [▶ Start Applying]
   ▼
Orchestrator (Claude Code session)
   ├── job-scout      → searches LinkedIn / SEEK / Indeed / company ATS boards
   ├── gatekeeper     → filters every job against YOUR criteria (never applies blindly)
   ├── cv-tailor      → rewrites your CV emphasis per job description, builds a PDF, QA-checks it
   ├── form-filler    → fills the application form, uploads the CV, submits
   └── digest-writer  → daily summary + application tracker

Everything is visible: a tracker file records every job (applied / skipped / why), the dashboard shows live pipeline stages, and every submission happens either in your own Chrome tab or is reported back with evidence.

Agent	Model	Job
`job-scout`	Sonnet (fast/cheap)	Searches sources, returns compact JSON per job
`gatekeeper`	Opus (judgment)	Runs your filters — terminal, no override
`cv-tailor`	Opus (judgment)	Re-emphasizes your CV per JD, builds + QA-checks the PDF
`form-filler`	Sonnet (fast/cheap)	Fills and submits authorized applications
`digest-writer`	Sonnet	Daily digest + tracker updates

Total setup time is about 15 minutes, most of it installing prerequisites.

The model split is deliberate: Opus handles the two steps that require judgment — filtering jobs against your rules and tailoring your CV without inventing anything — while Sonnet handles the fast, mechanical steps (searching, form-filling). Plain Python scripts do zero-token fetching wherever an API allows it. A typical full run costs a few cents of API usage per applied job.

What you'll need before you start

The template ships a checker script that verifies every one of these and prints the exact install command for anything missing — you don't have to check by hand.

Requirement	Why	Install
Claude Code (Pro/Max plan recommended)	Runs the agents	`npm install -g @anthropic-ai/claude-code`
Node.js 18+	Dashboard server (zero dependencies)	nodejs.org
Python 3.9+	CV builder + fetch scripts	python.org (macOS already has it)
`reportlab` + `pypdf`	PDF CV generation	`pip3 install reportlab pypdf`
`poppler` (`pdftoppm`)	CV visual QA	macOS: `brew install poppler` · Linux: `apt install poppler-utils`
playwright-cli	Fast headless form-filling on friendly ATS platforms	`npm install -g @playwright/cli`
Google Chrome + Claude in Chrome extension	Applying inside YOUR logged-in browser (LinkedIn, strict ATS)	Chrome Web Store → "Claude in Chrome" → install → sign in with your Claude account
Apify account + API key (optional)	Zero-token SEEK / job-board scraping	apify.com → free tier is enough

What this agent will never do

Solve or bypass CAPTCHAs and bot-checks
Create accounts or type passwords — it uses your already-logged-in browser sessions and never sees your credentials
Run LinkedIn unattended — you are always present, and it stops instantly on any security challenge
Fabricate CV content, skills or experience — re-emphasis of your real experience only
Apply to anything failing your filters — filter failures are terminal, there is no override
Submit anything you did not authorize

These rules live in CLAUDE.md, and every sub-agent reads them on every run. They are what keep your accounts safe and your applications honest.

Which job platforms it works with

The agent's ATS compatibility map, learned from real runs:

Platform	Agent behavior
Teamtailor, most public forms	Full auto via playwright-cli (headless, fast, cheap)
Ashby	Spam-flags headless browsers. Agent fills via Chrome extension in your real browser instead — proven to work
Cloudflare-fronted careers sites	CAPTCHA wall → agent prepares everything, you click submit
Workday	Requires an account → agent prepares everything, you submit (it will never create accounts)
LinkedIn Easy Apply	Supervised only, human-paced, instant stop on challenges
Indeed redirect links (`to.indeed.com`)	Never used — the agent finds the company's own ATS posting instead

The agent updates this map in its own memory as it learns new platforms.

What's inside the setup guide

Unlock the full step-by-step setup guide

1. Create the project folder
2. The master prompt — one paste builds everything
3. What just got created
4. Run /onboard-me (once)
5. Put your CV in
6. Connect Chrome (once per session)
7. Run /apply-jobs-automatically (every day)
8. Your daily routine — quick reference
9. Troubleshooting
Appendix A — Project structure
Appendix B — Credentials & what the agent can see
Appendix C — Plugins & discovery

Unlock the full setup guide

What's included

Your purchase includes the complete template file — every project file, byte-identical and ready to paste into a fresh Claude Code session: 5 sub-agent definitions, 3 slash commands, 4 scripts, the dashboard, and the rulebook that every agent reads on every run — plus the full numbered setup guide above and lifetime updates as the agent template improves. Nothing else to download, and no plugin registration steps: Claude Code discovers the agents and commands from their file paths automatically the moment you restart your session.

Frequently asked questions

Is this safe for my LinkedIn account?
Yes — LinkedIn automation runs supervised only, meaning you are present and watching every time. The agent stops instantly on any security challenge or CAPTCHA and never runs LinkedIn unattended, which is the behavior that gets accounts banned. It also never types your password or creates an account; it uses your own already-logged-in Chrome session via the Claude in Chrome extension.
Does it fabricate or exaggerate anything on my CV?
No. Your CV lives as a Python dict of your real experience, and the cv-tailor agent may only re-emphasize or reorder what's already in that dict per job description — it cannot invent skills or experience. You also give it a no-claim list of technologies you don't actually know, which it's forbidden from ever mentioning. Every generated PDF is rendered to images and visually QA-checked before use.
Do I need an API key to use this?
Only one optional key: an Apify API key for zero-token SEEK and job-board scraping, and the free tier is enough for most usage. Everything else runs on your existing Claude Code login — there's no separate LLM API key to manage, and the agent never reads any credential besides the optional Apify key in your local .env file.
What does a run cost?
A typical full run — search, filter, tailor a CV, fill and submit a form — costs a few cents of API usage per applied job. The pipeline is deliberately split so Opus (the more expensive model) only runs the two steps that need judgment — filtering and CV tailoring — while Sonnet handles the fast, mechanical steps like searching and form-filling, and plain Python scripts do zero-token fetching wherever possible.
Does it work on Windows and Linux, or just macOS?
The dashboard is a dependency-free Node.js server and the scripts are plain Python, so both run on any OS Node 18+ and Python 3.9+ support, including Windows and Linux. The one platform-specific prerequisite is poppler for CV visual QA — macOS installs it with brew install poppler, Linux with apt install poppler-utils.
What happens when a job application has a CAPTCHA?
The agent never solves or bypasses CAPTCHAs or bot-checks — that's a hard, non-negotiable rule. When it hits one, or any Cloudflare-fronted or account-required application (like Workday), it prepares everything it can and hands the job back to you on the dashboard for a manual submit, rather than trying to get around it.
What exactly do I get when I buy this?
One template document containing every project file — 5 sub-agent definitions, 3 slash commands, 4 scripts, the dashboard, and the rulebook — ready to paste into an empty Claude Code folder, plus the full numbered setup guide unlocked on this page and lifetime updates as the template improves. There's nothing else to install or subscribe to beyond the prerequisites listed above.
Do I need a paid Claude plan?
A Claude Code Pro or Max plan is recommended so you have enough usage headroom for daily runs, but the agent works on any plan that can run Claude Code and the Claude in Chrome extension. Actual per-run cost is small — typically a few cents of API usage per applied job — since only the judgment steps (filtering, CV tailoring) run on the more expensive model.

Ready to set it up?

The full numbered setup guide, plus lifetime updates as the agent improves.

Get Apply Jobs Automatically — AI Agent