If you have seen “Career-Ops” mentioned in Claude Code discussions, GitHub trending, or AI engineering Twitter and wondered what it actually is — this is the 30-second answer plus the context that makes the rest of the discussion legible.
The 30-Second Definition
Career-Ops is a Claude Code Skill bundle that turns the job search into an agentic pipeline. Instead of reading job descriptions one by one, hand-adapting your CV, and tracking applications in a spreadsheet, you point Claude Code at a list of job board URLs and Career-Ops handles the rest: scan, evaluate, score, generate a tailored PDF CV, log the state, and queue applications for your review.
The system was open-sourced in April 2026 by Santiago Fernández de Valderrama. The author used it on his own search: 740 evaluated listings → 66 applications submitted → 12 interview processes → 1 signed offer for a Head of Applied AI role.
Within a week the GitHub repository santifer/career-ops crossed 43,800 stars. As of May 2026 it sits at 44,554 stars and 9,353 forks, with multiple community variants.
Why Career-Ops Went Viral
Three reasons make Career-Ops more than “another job-search automation”:
- It is a real Claude Code Skill bundle, not a script. The architecture is 14 separate skill modes (
oferta,pdf,scan,batch,tracker,apply, etc.) that Claude Code loads on demand. Each mode is a markdown file inmodes/, with its own context. This is what “production-grade Skills design” looks like. - It ships an opinionated evaluation rubric. Career-Ops includes a 10-dimension A–F evaluation system (blocks A through G) that scores every job posting against the candidate profile. The rubric is built into the prompts, not bolted on. You can adapt it, but the default reflects months of refinement on real applications.
- The author published the numbers. Most viral AI tools claim to work; Career-Ops shipped with audited metrics from a real job search. The 6.6:1 evaluate-to-apply ratio and 12:1 application-to-interview ratio are reproducible because the criteria are encoded in the modes themselves.
What Career-Ops Is Not
- Not an auto-applier. Every application requires user approval. Career-Ops fills the form, but the user clicks submit.
- Not a SaaS. It runs locally on your machine via Claude Code. No subscription, no cloud account. Your data stays on your filesystem.
- Not a CV builder. It generates ATS-optimized PDFs per job description, but it depends on you supplying a real CV during onboarding.
- Not a magic interview prep tool. The
interview-prepmode helps you structure STAR stories from your CV, but you still have to do the practice.
Who Career-Ops Is For
Career-Ops makes sense if you tick at least two of these boxes:
- You are running an active job search (10+ applications in pipeline)
- You already use Claude Code daily (or are willing to learn it)
- You apply to multiple job boards (Greenhouse, Ashby, Lever, etc.) where the same role is posted in different places
- You want to keep human judgment at the gate, but offload the evaluation grunt work
- You are comfortable with the terminal and reading SKILL.md files
If you are looking for a one-click web app, this is not it. Career-Ops trades zero-config setup for full transparency and local control.
The 14 Skill Modes at a Glance
Career-Ops ships 14 distinct skill modes, each handling one piece of the pipeline:
| Category | Modes |
|---|---|
| Scanning | scan (portal scanner across 45+ companies), pipeline (URL queue) |
| Evaluation | oferta (single-job A–F score), batch (parallel evaluation) |
| CV / Application | pdf (ATS-optimized CV per JD), apply (Playwright form fill) |
| Tracking | tracker (state management), contacto (LinkedIn outreach) |
| Interview | interview-prep (STAR story builder), and several supporting modes |
For the complete catalog with what each mode does, when to use it, and how the modes chain together, see Career-Ops Skills: Complete Comparison of All 12 Modes (2026) and the Career-Ops Complete Reference.
How to Get Started
Three options depending on your goal:
1. Just want to try it on one job?
Install Career-Ops and run the oferta mode against a single URL. You get an A–F evaluation report and can decide if the system is worth investing in. Total time: ~10 minutes.
2. Want the full pipeline? Follow How to Use Career-Ops on Claude Code: From Zero to First Application in 30 Minutes. It walks through CV onboarding, portal scanning, batch evaluation, and submitting your first reviewed application end-to-end.
3. Want to understand the design first? Read the Career-Ops Complete Reference — the 14 modes, the A–G evaluation rubric, the multi-agent design, and how three community forks differ.
Career-Ops vs Other AI Job Tools
Several AI job-search tools shipped in 2025–2026. Career-Ops differs in two specific ways:
- It is open-source. Most competitors (LazyApply, JobScan AI, etc.) are SaaS with subscription pricing. Career-Ops is MIT-licensed code you own.
- It is Claude Code-native. Career-Ops is built as Claude Code Skills, not as a standalone product. If you already use Claude Code, the integration is zero-cost. If you do not, learning Claude Code first is the prerequisite.
The closest open-source analog is community forks of Career-Ops that adapt it to Codex CLI or Cursor. They differ in scope and supported CLIs — see our Complete Reference for the comparison.
Frequently Asked Questions
What is Career-Ops in one sentence?
Career-Ops is a Claude Code Skill bundle that turns job hunting into an agentic pipeline — scanning portals, evaluating listings with an A–F rubric, generating ATS-optimized PDFs, and tracking applications, all locally on your machine.
Is Career-Ops free?
Yes. Career-Ops is MIT-licensed open-source software you install locally. You need a Claude Code subscription to run the AI portions (Pro $20/mo or Max $100-$200/mo), but Career-Ops itself has no separate fee.
Does Career-Ops auto-submit job applications?
No. Every application requires user approval. Career-Ops fills the form via Playwright but always pauses for you to review and click submit. This is intentional — the system is designed as a filter, not a spray-and-pray tool.
How is Career-Ops different from LazyApply or JobScan?
LazyApply and JobScan are SaaS products with subscription pricing and closed source. Career-Ops is open-source code that runs locally on your machine via Claude Code. You own and adapt the prompts, the evaluation rubric, and the workflow.
Do I need to be a developer to use Career-Ops?
You need to be comfortable with the terminal, installing CLIs, and reading markdown SKILL.md files. You do not need to write code, but you should be willing to edit configuration files and understand Claude Code basics. If “open terminal, type command, read output” is unfamiliar, the learning curve will be steep.
What is the success rate of Career-Ops?
The author’s published numbers: 740 listings evaluated, 66 applications submitted, 12 interview processes opened, 1 signed offer (Head of Applied AI). That is a 12:1 application-to-interview ratio and a 12:1 interview-to-offer ratio on his specific search. Your numbers will vary based on your CV, the roles you target, and how aggressively you tune the evaluation rubric.
Where do I install Career-Ops?
The official repository is at github.com/santifer/career-ops. Follow the README installation instructions — typically git clone into your Claude Code skills directory (~/.claude/skills/career-ops/), then onboard your CV. For a step-by-step walkthrough, see How to Use Career-Ops.
Related Articles
- Career-Ops Complete Reference (14 Skill Modes, 10 Evaluation Dimensions, 3 Forks)
- How to Use Career-Ops on Claude Code: Zero to First Application in 30 Minutes
- Career-Ops Skills: Complete Comparison of All 12 Modes
- 15 Best Claude Code Skills You Should Install in 2026
- Claude Code Best Practices: The Complete Guide