auto-pipeline
Default flow when you paste a job description URL or raw text. Runs the A–H evaluation, generates the report, exports a tailored PDF, and writes a tracker entry in one run, stopping only when it needs an answer from you.
What it does
career-ops invokes auto-pipeline whenever you drop a JD URL or pasted text into the prompt. The mode reads the listing, scores it against the rubric (five dimensions plus a holistic global score), writes the evaluation to reports/, generates the tailored PDF via Playwright, and updates the tracker entry, all in one run. It stops to ask you only when the company is on your blacklist, or when the posting comes through an agency and it needs you to confirm which one.
When to use it
Use auto-pipeline when you want the full pipeline executed automatically. It is the default mode for new listings entering your search.
What a run looks like
Paste a job URL with no sub-command and auto-pipeline runs the whole chain end to end: it extracts the description, checks the posting is live, scores it, saves the report, exports a tailored PDF, and records the entry in your tracker.
> https://job-boards.greenhouse.io/notion/jobs/4523451008
Extracting JD with Playwright... done.
Liveness gate: posting is live.
Evaluation: 4.6/5 — legitimacy verified.
Saved reports/072-notion-applied-ai-engineer-2026-05-18.md
Generated the tailored PDF via Playwright.
Score >= 4.5 — drafted Section H application answers in the report.
Updated data/applications.md (Report and PDF marked done).Good to know
- Paste a URL and auto-pipeline handles extraction for you: Playwright first for single-page portals like Lever, Ashby, Greenhouse and Workday, then WebFetch for static pages, then WebSearch as a last resort.
- A liveness gate runs before the evaluation. If the link is dead or the page is an empty "position filled" shell, the mode stops there instead of scoring phantom content — and if the entry came from
data/pipeline.md, it marks that row closed. - A LinkedIn page whose description stays a loading skeleton is not a closed job, and not a job description either: auto-pipeline stops before the evaluation, the report, the PDF and any tracker write, and asks you to paste the job text or give an employer or ATS link for the same role.
- Draft application answers are written only when the score is 4.5 or higher, so low-fit roles do not spend tokens on copy you will not send.
- The saved report uses the oferta format, including the one-line verdict after Block G and the recruiter checks Block G runs when a recruiter brought you the role.
- In a CLI that lets the agent rename its own session, the session takes the posting's name, such as
VP Infrastructure, Acme Health, once the job description is read. Setsession_title: offinconfig/profile.ymlto keep your own session titles.
Gotchas
If you only want the evaluation without the PDF (faster, cheaper), invoke /career-ops oferta instead. auto-pipeline costs more tokens because it runs the full chain.
Related
Modes
The 15 user-invocable career-ops modes — evaluate, tailor, apply, track, and prep — each a markdown skill that runs in any supported AI coding CLI.
pipeline
Reads URLs queued in data/pipeline.md and runs the full auto-pipeline against each one. Use it when you collected ten or twenty listings during the day and want to crunch them in a single session.