DEV Community

Cover image for I got tired of AI resume spam and ghost jobs, so I built an open-source CLI agent for developers
Akash Kaintura
Akash Kaintura

Posted on

I got tired of AI resume spam and ghost jobs, so I built an open-source CLI agent for developers

#career #programming #javascript #opensource

Job hunting in 2026 is mathematically broken.

Recruiters use ATS bots to reject candidates in 6 seconds. Companies post ghost jobs that stay open for months. And the market is flooded with paid SaaS tools charging $39/month to blast recruiters with generic, hallucinated AI resumes.

I am a backend engineer. When something in my workflow is broken and repetitive, I build an engineering pipeline.

So over the last few months, I built Career-Ops — a 100% local, CLI-agnostic job search automation pipeline designed specifically for developers.

GitHub: https://github.com/UGilfoyle/career-ops

Here is a quick breakdown of how it works under the hood.

1. Zero-Token Direct ATS Scraping

Most job scrapers either run heavy Playwright instances that consume 300MB+ RAM or burn expensive LLM tokens just parsing HTML.

In Career-Ops, we hit the underlying JSON endpoints of major ATS platforms directly:

  • Greenhouse (boards-api.greenhouse.io)
  • Ashby (api.ashbyhq.com)
  • Lever (api.lever.co)

This returns structured metadata in milliseconds with zero token cost.

For bot-protected portals (like Naukri or Indeed), we integrated a dual-engine architecture:

  • Primary: An Obscura CDP Daemon (~30MB RAM footprint with hardware fingerprint spoofing).
  • Fallback: Playwright Chromium with automated evasions.

2. Anti-Hallucination Resume Engine

Most AI resume builders fabricate skills you’ve never touched, leading to awkward interview rejections.

Career-Ops treats your genuine career experience (cv.md) as the immutable source of truth:

  • Dynamic Skill Reordering: If the JD targets DevOps, Cloud and Docker bullets bubble up. If it targets Distributed Systems, Redis and telemetry scale take priority.
  • Proof-Point Matching: Automatically connects your genuine STAR stories to the job requirements.
  • Zero Hallucination: Strict keyword budget without making up fake tools.

Outputs are compiled locally into pixel-perfect ATS-compliant HTML/PDF and LaTeX.

3. Local Privacy & Data Ownership

Your job search data, target salary, and interview prep shouldn't live on a 3rd-party SaaS server.

Everything in Career-Ops lives in local Markdown and TSV files on your laptop:

  • data/applications.md — Application tracking
  • data/pipeline.md — Job queue
  • interview-prep/story-bank.md — Accumulated STAR+R behavioral stories

You can version-control your entire job hunt with private Git commits.

4. CLI-Agnostic

Career-Ops follows the Open Agent Skill Standard. You can run it with whichever AI coding tool you prefer:

  • Claude Code (claude)
  • OpenCode (opencode)
  • Antigravity CLI (agy)
  • Codex / Gemini CLI

Try it out!

Career-Ops is completely free and open-source under the MIT license:

👉 GitHub Repository: https://github.com/UGilfoyle/career-ops

Would love to hear from fellow devs:

  1. What is the most frustrating part of your current job search?
  2. Which job boards or ATS platforms should we add scrapers for next?

Drop a star on GitHub if you find it helpful, or share your thoughts in the comments!

Top comments (0)