BuyerAware

Measured conclusion: Deep-tested. L2 outcome: tools_ok.

Mechanic-grade used-car listing verdicts: risk score, failure points, repair costs, fair price.

Add to your agent

Copy this into your MCP client config:

{
  "mcpServers": {
    "BuyerAware": {
      "url": "https://api.buyerawareapp.com/mcp",
      "transport": "streamable-http"
    }
  }
}

Measured tools (5)

• create_trial_key

  Create a BuyerAware API key so this user can run verdicts (3 free). Call this once when no key is configured, then show the user the key and tell them to save it. Do not call it again for someone who already has a key — one trial per email address.

  Parameters:

  • email (required) — string — the user's email address; the key is issued against it, one trial key per address
  • name — string — optional display name for the key record

  Hints: read-only no · destructive no · idempotent no · open-world no

• analyze_listing

  Start a BuyerAware pre-purchase verdict on ONE used vehicle. Runs a mechanic-grade read: ranked failure points with $ exposure, recalls, seller-claim fact-check, fair price range, adjusted true cost, risk score 1-10, BUY / BUY_WITH_CONDITIONS / WALK_AWAY, a word-for-word negotiation script, and lower-risk 'Tank' alternatives. Supply whichever of listing_url, listing_text or vin you have — at least one is required, and more context sharpens the verdict. Takes ~1-2 minutes: returns a job_id, then poll get_verdict. Costs 1 credit. For 'what should I buy instead' use tank_alternatives, which is free.

  Parameters:

  • vin — string — 17-character VIN, if the listing shows one
  • location — string — buyer city/state, for pricing and inspection context
  • listing_url — string — link to the listing — Marketplace, Craigslist, AutoTrader, a dealer page
  • listing_text — string — the listing copy pasted as-is, when there is no shareable URL
  • buyer_context — string — what the buyer needs it for, budget, and any specific concerns

  Hints: read-only no · destructive no · idempotent no · open-world yes

• get_verdict read-only

  Fetch the result of an analyze_listing job. If status is 'queued' or 'running', wait ~15 seconds and call again. Free — polling never costs a credit.

  Parameters:

  • job_id (required) — string — the job_id returned by analyze_listing

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• tank_alternatives read-only

  FREE, no API key needed, answers instantly. Recommend 2-3 lower-risk used vehicles ('Tanks') from BuyerAware's curated 47-platform database that land near a budget. Use when a buyer asks what to buy instead, or what is reliable for $X. Use analyze_listing instead when they have a specific car in mind.

  Parameters:

  • budget (required) — number — target purchase price in USD
  • limit — number — how many to return; default 3
  • mileage — number — mileage to price at; default 100000
  • category — string — sedan, hatchback, small suv, medium suv, large suv, midsize truck, full-size truck, diesel truck, minivan, wagon, hybrid, phev, ev, luxury
  • max_risk — number — risk ceiling 1-10; default 5

  Hints: read-only yes · destructive no · idempotent yes · open-world no

• tank_lookup read-only

  FREE, no API key needed, answers instantly. Look up one platform in BuyerAware's curated database: risk score, verdict, the one known trap, private-party value at a mileage, 5-year operating cost. Covers 47 US-market platforms; anything outside it returns no match, which is itself a signal — follow up with analyze_listing.

  Parameters:

  • query (required) — string — platform name, e.g. '4Runner', 'Camry V6', 'CR-V', 'Tundra 5.7'
  • mileage — number — mileage to value at; default 100000

  Hints: read-only yes · destructive no · idempotent yes · open-world no

Publisher website

Measured by MCPYP probing. Untested fields are omitted, never estimated. Listing is always free; MCPYP never proxies calls or stores user tokens.