Is this car a good deal?
A used car is a fair deal when its asking price sits inside the expected band from comparable dealer listings, and you can check that by asking Claude or ChatGPT to call DealScore.
What score_deal returns
score_deal returns these fields.
| Field | Meaning |
|---|---|
| grade | Great, Good, Fair, High, or Overpriced. Null when coverage is cannot_score. |
| score | 0 to 100. Each grade has its own range, so the score always matches the grade. Null when coverage is cannot_score. |
| expected_price | Market value estimate from comps, in USD. |
| expected_price_low | Low end of the comp price band, in USD. |
| expected_price_high | High end of the comp price band, in USD. |
| delta_pct | (asking price minus expected price) divided by expected price. |
| confidence | 0 to 1, from comp count and data source. |
| coverage | scored, partial, or cannot_score. Only scored is billed. |
| comps_count | How many comparable dealer listings the score used. |
| rubric_version | Which rubric mapped delta_pct to a grade. The current rubric is score_v2. |
| data_source | live, mock, or error on real responses, and "example" only on the illustrative example shown on this site. |
Example output, illustrative numbers
A generic 2019 sedan, asking $20,000. These numbers are round and made up.
{
"grade": "Good",
"score": 87,
"expected_price": 22000,
"expected_price_low": 20000,
"expected_price_high": 24000,
"delta_pct": -0.09,
"comps_count": 20,
"data_source": "example"
}Those figures are illustrative, and numbers for a real listing come from a live, per-request score. Score a real listing.
score_v2 grade bands
Rubric version score_v2 maps delta_pct to a grade and a score. delta_pct is (asking price minus expected price) divided by expected price. Each grade has its own score range: Great 90 to 100, Good 75 to 89, Fair 60 to 74, High 40 to 59, Overpriced 0 to 39. Inside a range, a lower price scores higher, and only a price 20% or more under expected scores 100.
| Grade | delta_pct | Score | Meaning |
|---|---|---|---|
| Great | ≤ -10% | 90 to 100 | Asking well below expected market value. |
| Good | ≤ -3% | 75 to 89 | Asking modestly below market. |
| Fair | ≤ +3% | 60 to 74 | Roughly in line with comps. |
| High | ≤ +10% | 40 to 59 | Asking above market, but not wildly so. |
| Overpriced | > +10% | 0 to 39 | Asking well above what comps support. |
Set up Claude
- Connect DealScore in Claude Desktop. Full steps are on /integrations/claude-desktop.
- Ask Claude to call score_deal with the year, make, model, asking price, and mileage.
Set up ChatGPT
- Turn on Developer mode and add the DealScore MCP URL. Full steps are on /integrations/chatgpt.
- Ask ChatGPT to call score_deal with the year, make, model, asking price, and mileage.
Cursor, VS Code, Claude Code, Codex, and n8n use the same MCP server. Each client has a setup page. Cursor, VS Code, Claude Code, Claude Desktop, Codex, ChatGPT, n8n.
FAQ
- How do I know if a used car price is fair?
- A used car price is fair when the asking price sits inside the expected band from comparable dealer listings. Ask Claude or ChatGPT to call DealScore. score_deal returns a grade, a 0 to 100 score, and that band.
- What does a Fair grade mean?
- Fair means delta_pct is greater than -3% and at most +3%. The asking price is roughly in line with comps.
- Does DealScore use KBB or Edmunds values?
- No. DealScore does not use KBB or Edmunds values. Comps are current dealer listings from a licensed listings data provider. If you send a trim, the score does not use a different trim. The default comps are national. A ZIP narrows them to a local market.
- Can I check a private-party listing?
- Yes. Paste the listing text into the chat. The agent reads the year, make, model, asking price, and mileage, then calls score_deal. DealScore does not fetch Facebook Marketplace or Craigslist URLs. Comps are US dealer listings, so a private-party price is judged against dealer prices for comparable cars.
- Is it free?
- Free is 5 scored calls a month. No card. Builder is $19/mo (founding price) for 1,000 scored calls a month.
Get a free API key
Get a free API key, or connect the same account over MCP.
Connect over MCP