career-ops: Candidates Finally Get the AI Arms Race Going Their Way
Companies have used AI to filter job applicants for a decade. career-ops flips the table: an open-source job-search command center that runs inside the agent CLI you already use, Claude Code, Codex, OpenCode or Antigravity, and puts the machine on the candidate's side. It has been on GitHub's trending board three separate days now, added 855 stars today alone, and sits at 66,600 total with WIRED and Business Insider coverage. The creator's line is the whole pitch: companies use AI to filter candidates, I just gave candidates AI to choose companies.
The design is opinionated in the right places. An A-through-H evaluation system scores every listing 1.0 to 5.0 across role fit, compensation research and legitimacy, with a dedicated block for flagging scams and ghost jobs. The agent navigates job portals with Playwright, reasons about fit rather than keyword-matching, generates ATS-optimized PDFs per listing, drafts cover letters and recruiter outreach, and tracks the whole pipeline in a terminal dashboard. Two rules keep it honest: nothing ever auto-submits, drafts only, and the docs tell you to skip anything scoring under 4.0 out of 5 rather than spray-and-pray. The author reports evaluating 740+ listings with it and landing a Head of Applied AI role.
The interesting tension: if every candidate runs this, the ATS-versus-tailored-resume game becomes fully machine-versus-machine and the signal in an application drops to zero. That is not an argument against career-ops, it is an argument that the AI-filtered hiring pipeline was already broken, and arming both sides equally is how broken equilibria get renegotiated. MIT licensed.
https://github.com/santifer/career-ops
← Back to all articles
The design is opinionated in the right places. An A-through-H evaluation system scores every listing 1.0 to 5.0 across role fit, compensation research and legitimacy, with a dedicated block for flagging scams and ghost jobs. The agent navigates job portals with Playwright, reasons about fit rather than keyword-matching, generates ATS-optimized PDFs per listing, drafts cover letters and recruiter outreach, and tracks the whole pipeline in a terminal dashboard. Two rules keep it honest: nothing ever auto-submits, drafts only, and the docs tell you to skip anything scoring under 4.0 out of 5 rather than spray-and-pray. The author reports evaluating 740+ listings with it and landing a Head of Applied AI role.
The interesting tension: if every candidate runs this, the ATS-versus-tailored-resume game becomes fully machine-versus-machine and the signal in an application drops to zero. That is not an argument against career-ops, it is an argument that the AI-filtered hiring pipeline was already broken, and arming both sides equally is how broken equilibria get renegotiated. MIT licensed.
https://github.com/santifer/career-ops
Comments