job-searchlisted

# Job search scan A search that runs on a schedule beats one you remember to do. This skill turns a scattered search into one repeatable pass: same sources, same rubric, same report shape, diffed against last time so you only read what changed. **Everything personal lives in `config.yml`, not in this skill.** Copy `config.example.yml` to `config.yml`, fill it in, and never commit it. This skill ships with no profile, no employer names, and no target list. ## Setup (once) 1. Copy `config.example.yml` to `config.yml` in this skill's directory. 2. Fill in: your tracks, must-haves, sources, excluded companies, and where reports get written. 3. Connect whichever integrations you want — see `references/integrations.md`. None are required; the scan degrades to whatever is available and says so in the report. ## Workflow 1. **Load config.** Read `config.yml`. If it is missing, stop and say so — do not guess a profile or invent target companies. 2. **Scan sources** (`references/sources.md`). Work through each configured source. Prefer a company's own ATS posting over an aggregator — aggregator URLs rot and often lag the real board. 3. **Sweep email**, if configured. Job alerts and direct recruiter outreach. Recruiter mail outranks any board listing: a human already thinks you fit. Read-only — never reply or label. 4. **Check community channels**, if configured. Roles get posted in Slack and Discord days before they hit a board, and some never hit one at all. 5