Skill: lead-harvest

Goal

Harvest raw lead candidates from competitor references + industry directories without violating ToS.

Inputs

• •config/competitors.yaml
• •config/sources.yaml
• •policies.yaml

Outputs

• •staging/leads_raw.parquet
• •logs/harvest.log

Procedure

1. •For each competitor domain:
   • •discover pages with keywords: kunden|referenzen|case study|success story
   • •crawl depth=2 (configurable)
2. •Extract candidate entities:
   • •company name, location, website (if present), evidence_url, snippet
3. •For each directory/fair source:
   • •run the corresponding spider / extractor
4. •Store each lead with:
   • •evidence_url, fetched_at, content_hash