career-ops (santifer)

One-line summary: Open-source AI job search system built as a set of Claude Code skills; scans career pages, rewrites CVs per job, and fills application forms — the tool Paul is currently using (as a submodule in pmcclelland/career-development) for Track 1 job-search execution.

What it is

A GitHub-hosted automation toolkit for AI-driven job search (github.com/santifer/career-ops), built on Claude Code. Described in 2026-04-20-job-search-tool as having:

• 14 skill modes (evaluate, scan, PDF, etc.)
• Go terminal dashboard for managing the search queue
• ATS-optimized PDF generation via Playwright
• 45+ companies pre-configured (Anthropic, OpenAI, ElevenLabs, Stripe, and others)
• Capabilities: scans multiple company career pages, rewrites CV per job, fills application forms
• Bilingual Spanish README (per reply thread)

The author (@santifer, surfaced by @Hesamation on X) reports using it for 700+ job applications, which produced a hire — the single data point behind the tool's visibility.

Why it matters to career

Three reasons:

1. It's the execution tooling for Track 1. Paul's pmcclelland/career-development repo already includes career-ops as a submodule with upstream tracked — see the 2026-04-17 execution repo note in the log. This entity documents the upstream tool itself.
2. Itself a portfolio signal. A reply in the thread ([@yoemsri in 2026-04-20-job-search-tool]): "if I'm a hiring manager and I see he built this, I'm hiring him on the spot just for the engineering alone." Thin evidence (one hiring-manager-perspective reply), but directionally aligned with what what-makes-compelling-frontend-portfolio-for-ai-era already concludes: building public AI tooling is itself the credential.
3. Single working-agent-loop data point. From @OpenGPUnetwork in 2026-04-20-job-search-tool: "That's not a demo, that's a working agent loop with measurable output." One of the more concrete published instances of a Claude Code-native multi-skill agent loop producing real-world outcomes — relevant to ai-native-multi-agent-workflow.

Key facts

• Creator: @santifer
• License / model: open source (MIT or similar — confirm from repo)
• Platform: Claude Code with 14 skill modes
• Stack: Go (terminal UI), Playwright (PDF generation)
• Surfaced by: @Hesamation X thread, 2026-04-05

Strengths (from career's perspective)

• Ready-made Track 1 execution layer — avoids building from scratch; the existing structure (skills, dashboard, company config, ATS PDF) covers the main mechanical surface of a job search
• Active prior art — the tool has a concrete success case and a public discussion, lowering the risk of adopting an abandoned side project
• Forkable — Paul's private copy sits under his control, so customizations (writing voice, target-company list, filtering heuristics) don't require upstream cooperation

Weaknesses (from career's perspective)

• Race-to-the-bottom dynamics. As the tool becomes widely adopted, its marginal advantage decays — see ai-job-application-arms-race for the dynamic. Volume-based automated applying becomes less effective as employers defend (no public JDs, anti-AI prompt injections, live-interview gating).
• Referral channel unaffected. 90% of senior hires come from referrals, not inbound (how-competitive-is-senior-frontend-job-market) — this tool only optimizes the 10% inbound channel.
• Blast radius if weaponized carelessly. Mass-applying to 700 companies produces a visible digital footprint; selective, high-quality targeting likely outperforms volume for senior/principal roles where reviewers are human and attentive.
• No evidence it's tuned for principal/senior roles. The success case is one hire; the target level and company type aren't disclosed in the thread.

Open questions

• Does career-ops actually meaningfully improve outcomes at senior/principal level, where hiring is judgment-driven and applications are read by humans?
• How does the tool handle company-specific application processes that don't fit a generic form-fill pattern?
• What's the update cadence — does it stay current with Claude Code skill evolution, or is it frozen?
• Are employers detecting and filtering career-ops-generated applications specifically (pattern detection)?

Sources

• 2026-04-20-job-search-tool

Related

• ai-job-application-arms-race — the broader dynamic career-ops both exploits and accelerates
• ai-native-multi-agent-workflow — career-ops is one public example of a Claude Code multi-skill agent loop
• how-competitive-is-senior-frontend-job-market — the search conditions career-ops is deployed into
• what-makes-compelling-frontend-portfolio-for-ai-era — the "building this is itself the portfolio" signal