Career-Ops

What is Career-Ops?

Career-Ops is an open-source, agentic job search system that turns an AI coding CLI into a “command center” for researching roles, evaluating fit, generating tailored CVs as PDFs, and tracking applications in a single workflow. It’s designed to help candidates manage the end-to-end process more systematically than manually maintaining spreadsheets or sending the same application to many roles.

The system uses Claude Code to navigate career pages, evaluate offers using a structured scoring rubric, and adapt your resume per job description. It’s positioned as a filter rather than a spray-and-pray tool, including recommendations to avoid applying to offers that score below a threshold (4.0/5) and emphasizing that the system never submits applications on your behalf.

Key Features

• Auto-pipeline from a URL: Provide a job or company URL to generate an evaluation, create a PDF CV, and add a tracker entry for that role.
• Structured A–F evaluation with weighted dimensions: Uses a 6-block evaluation approach (role summary, CV match, level strategy, comp research, personalization, interview prep) grounded in your CV versus the job description rather than keyword matching.
• Tailored ATS-optimized PDF generation: Produces job-specific CV PDFs using design references (“Space Grotesk + DM Sans”) while injecting the relevant keywords/content for each job description.
• Portal scanning and scraping support: Includes a “portal scanner” with pre-configured company portals and the ability to customize additional queries; it’s explicitly described as covering portals such as Greenhouse, Ashby, Lever, and company career pages.
• Batch processing with parallel workers: Evaluates “10+ offers in parallel” using claude -p workers (sub-agent behavior), then merges results into a consistent pipeline.
• Terminal dashboard (TUI): A text-based dashboard to browse, filter, and sort the job pipeline.
• Pipeline integrity and human-in-the-loop control: Includes automated merge/deduplication, status normalization, and health checks while keeping the final decision and actions with the user (no application submission automation).

How to Use Career-Ops

1. Clone and install the repository, then install dependencies for browser automation used in portal navigation and PDF generation (via Playwright, including Chromium).
2. Run the prerequisite check with npm run doctor to validate setup.
3. Configure profiles and portals by copying example YAML files to config/profile.yml and portals.yml, then editing with your details and the companies/portals you want to scan.
4. Add your CV by creating cv.md in the repository root (CV content is provided in Markdown).
5. Personalize the system using Claude Code in the project directory. The source suggests updating modes/archetypes, translating modes, adding companies to portals.yml, and incorporating your CV into the system context.
6. Run evaluations and review recommendations through the pipeline/dashboard. The workflow is explicitly “human-in-the-loop”: the system evaluates and recommends, and you review before submitting anything.

Use Cases

• Filter a large set of openings for fit: When you have many job links, run batch evaluations in parallel, then use the dashboard to focus only on roles with sufficiently high scores (the system recommends avoiding anything below 4.0/5).
• Generate one CV per role with ATS-oriented formatting: For each job description, create a tailored ATS-optimized PDF CV rather than maintaining a single static resume.
• Prepare for behavioral interviews using accumulated stories: Use the “Interview Story Bank,” which accumulates STAR+Reflection stories across evaluations so you can answer behavioral questions using a small set of master stories.
• Scan known applicant tracking ecosystems efficiently: Evaluate roles across multiple hiring portals by using the configured portal scanner for platforms such as Greenhouse, Ashby, and Lever, plus custom queries.
• Maintain a single source of truth across a pipeline: Let the system track evaluations and statuses with integrity checks (deduplication and status normalization) so you can browse and manage progress from one place.

FAQ

Is Career-Ops a spray-and-pray application tool? No. The source describes Career-Ops as a filter designed to help you identify the small set of offers worth your time. It also recommends against applying to roles scoring below 4.0/5, and you are responsible for reviewing before submitting.

Does Career-Ops submit applications automatically? No. The system is described as evaluating and recommending, with “Human-in-the-Loop” control—your final action is required, and the system never submits an application.

What does it use to navigate career pages? The system uses Claude Code with Playwright to navigate career pages.

How are CVs and outputs produced? Career-Ops generates tailored PDFs (job-specific CVs described as ATS-optimized) and maintains pipeline tracking entries while processing jobs in batch.

How do you provide your CV to the system? The quick start indicates creating cv.md in the project root with your CV written in Markdown.

Alternatives

• Manual spreadsheet + templated resume workflow: A non-agent approach where you track applications in a sheet and manually tailor a resume. Compared to Career-Ops, it lacks automated evaluation, batch portal scanning, and structured scoring.
• Generic job-agency aggregators with resume upload: Tools that collect openings and assist with applying, typically focusing on matching and application management. Career-Ops differs by emphasizing an agentic evaluation process (CV vs job description) and per-job PDF CV generation.
• AI resume builders without job-specific evaluation: Services that generate resumes from prompts but don’t perform a structured, job-by-job fit evaluation with a scoring rubric and pipeline integrity features.
• Custom scraping + tracker + LLM scripts: A developer-built pipeline that can scan portals and summarize results. Career-Ops bundles a particular workflow (scoring rubric, dashboard, PDF generation, batch/sub-agent processing) rather than leaving all integration work to you.