Documentation Index
Fetch the complete documentation index at: https://mintlify.com/MadsLorentzen/ai-job-search/llms.txt
Use this file to discover all available pages before exploring further.
/rank sits between /scrape and /apply. After a scrape returns more postings than you want to evaluate by eye, /rank batch-scores every new one against your fit framework using parallel agents — each agent fetches the live posting and scores it across four dimensions — then returns a ranked shortlist with honest per-job strengths, gaps, deal-breaker vetoes, and deadline urgency flags. The scores are deliberately lightweight triage, not final evaluations: /apply always re-evaluates with full company research when you decide to apply.
Usage
/rank scores only postings with status new in job_scraper/seen_jobs.json. Jobs that already appear in job_search_tracker.csv (applied to or consciously tracked) are excluded regardless of flags. The --all flag re-scores every non-applied posting, including ones ranked in previous runs.
How Scoring Works
/rank dispatches parallel general-purpose agents via the Agent tool, roughly 5 jobs per agent. Each agent receives the job list, the posting URLs, and a compact scoring rubric extracted from your profile files — it never re-reads those files from disk. Each agent fetches its assigned posting URLs with WebFetch and scores only from the content it actually retrieves.
Each job is scored across four dimensions drawn from your 04-job-evaluation.md framework:
| Dimension | Weight | What it measures |
|---|---|---|
| Technical | 30% | How well the role’s required and preferred skills match your technical profile |
| Experience | 25% | How directly your work history maps to the responsibilities described in the posting |
| Behavioral | 15% | How well the role’s culture and work style align with your behavioral profile |
| Career Alignment | 30% | How well the role fits your target directions, motivators, and stated career goals |
- Location:
PASS(within acceptable commute),FAIL(requires relocation or is definitively outside your range), orFLAG(heavy travel or borderline distance — stays in ranking but marked with ⚠ for your judgment). - Deadline urgency: postings with a deadline within 7 days receive a 🔥 marker and win ties in the ranking. Postings whose deadline has already passed are moved to
expiredwithout being scored.
FAIL excludes a job from the shortlist entirely, no matter how high its content score. Vetoed jobs are listed separately with the reason.
Expired posting detection: if an agent cannot fetch a posting URL, or the page has expired or redirected to a listing index, the job is marked expired — it is never scored from the title alone and content is never fabricated.
The overall score is computed as a weighted average of the four dimensions. Jobs are then mapped to verdict bands: Strong Fit (75+), Good Fit (60–74), Moderate Fit (45–59), Weak Fit (30–44), Poor Fit (below 30).
Triage vs. Full Evaluation
/rank scores from posting text and your profile only. It does not perform company research, salary lookups, or spawn a reviewer agent — that depth belongs to /apply and would make /rank too expensive to run on every scrape batch. The triage scores are a routing decision, not a hiring decision. /apply’s Step 1 evaluation, which adds live company research and can incorporate salary benchmarks, always re-runs from scratch when you choose to apply to a job — triage scores are passed as prior context but never substitute for the full evaluation.Reading the Output
/rank asks which jobs you want to apply to. Give it one or more numbers and it hands off to the full /apply workflow on those posting URLs, passing the triage verdict as prior context.
Keeping Scores Fresh
After you update your profile — adding a new skill, changing your career goals, or running/setup --section experience — the scores from a previous /rank run may no longer reflect your current fit. Use --all to re-score everything:
seen_jobs.json, including ones that were already ranked. State is updated in place: status, rank_score, rank_verdict, and rank_date fields are overwritten for re-scored jobs. The underlying deduplication structure used by /scrape is never modified, so re-ranking never interferes with future scrape runs.
For more on what happens after you pick a job to apply to, see the /apply command guide. To get new postings to rank, see the /scrape command guide.