vac-company-researchlisted

# VAC Company Research Research a company automatically to generate the intake briefing for voice agent creation. This replaces the human intake step in the Auto Voice Pipeline. **Dual role:** (1) auto-intake bootstrap (Step 0). (2) **Company-detail source for the KB** — invoked by `vac-knowledge-base` whenever a company-info KB is needed, even on the human-intake path. The split mirrors the KB boundary: **the few most-asked company facts + the call purpose → prompt (tier-1); detailed company questions → KB (tier-2).** When a KB is in play, do the deeper Step 4b pass below. **Language note:** this skill is instruction text in English, but its output artifact `research-company.md` keeps German field labels and German industry values — it feeds a German-speaking voice agent, and downstream skills read those labels. ## Input You receive a spawn prompt with: - `firmenname`: Company name - `website`: Company URL/domain - `telefon`: Phone number (for the demo call) - `ansprechpartner`: Contact person name (optional) - `email`: Contact email (optional) ## Workflow ### Step 1: Website Scraping Scrape the company website systematically: ``` 1. Homepage → elevator pitch, Branche, Tonalitaet 2. /ueber-uns OR /about → company history, team, values 3. /leistungen OR /produkte OR /services → core products/services 4. /karriere OR /jobs → hiring signals, team size 5. /impressum → legal form, location, managing directors ``` Use `web_fetch` for each page. If a page 404s, skip it. U