broker-price: Authenticated Price Research
Searches prices across all bound platforms using the user's own account session — the same search they'd do manually in each app. No web scraping, no anti-bot issues.
Workflow
- Receive product info (from broker-recognize or user description)
- Build platform-specific search queries
- For each bound platform with valid session:
- Search active listings → extract prices, conditions
- Search sold listings → extract actual transaction prices
- Calculate:
- Price range (min → max)
- Median active listing price
- Median sold price (preferred metric)
- Market sentiment (seller-favorable / balanced / buyer-favorable)
- Recommend price: median of sold comparables (when available), or 5% below active median (when no sold data)
- Present transparency report with data sources
Script
python3 scripts/research.py --product product.json
# Searches all bound platforms, outputs transparency report
Fallback
If session-based search fails (expired session, platform API change), fall back to Claude Code's WebSearch for comparable listings. Results from fallback are clearly labeled as lower-confidence estimates.