
AI agents
Classic car data for AI agents
Old Cars Data exposes collector car auction history, classified inventory and sale history, live auctions, bid trails, and market statistics through REST and MCP so agents can research inventory, asking prices, comparable sales, and market activity.
Page updated
Based on Old Cars Data coverage from Bring a Trailer, Cars & Bids, Hemmings, AutoHunter, Hagerty, PCAR Market, All Collector Cars, Gooding & Co, RM Sotheby's, Barrett-Jackson, duPont REGISTRY Live, Sotheby's Motorsport, Car & Classic, The Market, Broad Arrow, MB Market, Mecum Auctions, PistonHeads, Collecting Cars, Bonhams.
What agents can ask
- What Porsche 911s sold recently, and for how much?
- Which Toyota Land Cruiser live auctions are ending soon?
- What are comparable sales for this VIN, seller, make, model, or year range?
- How does current live bidding compare with completed sales?
- What is the median sale price and sell-through context for this market?
- Which makes and models are available before the agent runs a filtered search?
- Which dealer listings match this vehicle, and how do asking prices compare with confirmed sold prices?
Why MCP matters
MCP gives AI tools a typed interface to current collector-car data. Instead of relying on a model memory, scraping, or a brittle browser workflow, the agent can call purpose-built tools for normalized makes, models, auctions, live listings, bid trails, and statistics.
Classified inventory and sale history
Use search_classifieds or GET /classifieds for current inventory and retained classified history. Both accept the same filters and share cursor pagination and the monthly query allowance. Sold history includes listings whose source pages disappeared.
Keep asking prices, dated asking_price_history, monthly financing, and confirmed sold prices separate. Confirmed missing adverts become gone without implying a sale; known sold and withdrawn outcomes remain intact. Use last_seen_at for freshness and updated_at for syncing stored updates, including routine checks. Use changed_since for changes to listing facts without routine freshness updates, and last_seen_before for an exclusive stale-presence cutoff. changed_since orders by changed_at and id; unknown last_seen_at values do not match last_seen_before. Successful pages still count toward the existing query allowance. Completed changed_since polls in asc order return meta.next_sync_cursor; pass it as cursor on later polls with the original filters. Completed checkpoints have no time expiry, while page cursors still expire after 24 hours. Normal full refreshes preserve unchanged content clocks. Use status=all with updated_since for a complete stored change feed, keep filters fixed while paging, overlap polling windows, and upsert by ID.
MCP tool map
| Tool | Agent job |
|---|---|
| list_makes | Discover available makes and avoid invalid query names |
| list_models | Find canonical model names for a selected make |
| search_auctions | Pull completed sales, vehicle details, source URLs, and auction outcomes |
| search_classifieds | Search dealer inventory, asking prices, and retained classified sale history with cursors |
| search_live_auctions | Monitor active listings by make, model, price, year, bids, and ending window |
| get_auction_bids | Inspect bid progression and demand behavior |
| get_price_stats | Summarize market statistics and recent trend context |
Source citations
Old Cars Data preserves source-platform context and source URLs where available, then normalizes those records into REST, MCP, model pages, and markdown answer pages.
Frequently asked questions
Can AI agents use Old Cars Data directly?
Yes. Old Cars Data exposes a remote MCP server so compatible AI clients can query collector car makes, models, auction history, classified inventory and sale history, live auctions, bid trails, and statistics.
Why use MCP instead of pasting search results into a chat?
MCP lets the AI client call structured tools against current data. That is more reliable than asking a model to infer values from stale or unstructured text.
What kinds of agent workflows fit Old Cars Data?
Good fits include comparable-sale research, live-auction monitoring, bid-limit support, dealer inventory enrichment, appraisal research, and market trend summaries.