career driven

Managing a multi-lane job search across dozens of disparate job boards and careers sites is a manual, fragmented mess. This skill localizes your entire search pipeline, providing a command-line engine to ingest your history, crawl niche job boards, rank leads by your specific criteria, and generate tailored, submission-ready PDFs.

What it does

• Local fact banking maintains a single source of truth for your professional history in content/ to prevent fact invention.
• Automated job discovery crawls Greenhouse, Lever, Ashby, Workday, and other public APIs to aggregate new leads.
• Lead ranking scores job postings against your profile.yaml keywords, target titles, and lane signals.
• Tailored document generation renders role-specific resumes and cover letters using Typst for high-quality PDFs and Word for ATS compatibility.
• Application tracking logs every sent document and interview stage in a structured local file system.

How it works

1. Initialize the system by running career init and career onboard ingest to transcribe existing resumes into the local fact bank.
2. Configure your search by defining target companies and job board watchers in watch.yaml and setting ranking signals in profile.yaml.
3. Run discovery with career discover run to fetch new postings and update your REVIEW-QUEUE.md.
4. Tailor and promote leads using career build to generate a job-specific overlay and draft your application materials.

Frameworks & tools

The engine is a uv script requiring Python dependencies like pyyaml and python-docx. It uses typst for PDF rendering, poppler for PDF parsing, and optional playwright for HAR-based site discovery.

Why this beats prompting it yourself

Standard LLM prompting often results in "hallucinated" achievements or generic buzzwords that fail ATS screens. This skill enforces a "content-first" rule where every resume line must map back to a verified fact bank, ensuring 100% accuracy while automating the repetitive crawl-and-format loop.

Use cases

• Manage high-volume applications across multiple career "lanes" like Product Management and Engineering.
• Fix broken job board integrations by importing browser HAR files to bypass fingerprinting issues.
• Maintain a permanent, immutable record of exactly which resume version was sent to which company.
• Prepare for interviews by instantly retrieving facts and notes associated with a specific application folder.

Known limitations

Does not support one-off resume advice or OCR for image-only PDFs. Requires local installation of uv, Typst, and poppler. Cannot bypass robots.txt or site-wide bot screening.