# Used Car Market Comparison > Compare used-car asking prices with matched local listings through Crawlora. Use for a vehicle shortlist, a comparable-listing price range, or explaining a listing's premium or discount by trim, mileage, condition, and location. - Skill: `crawlora-org/used-car-market-comparison` (Agent Skill, multi-file: 3 files) - Install (CLI): `npx skillmds@latest add crawlora-org/used-car-market-comparison` - Raw SKILL.md: https://api.skillmd.com/api/skills/crawlora-org/used-car-market-comparison/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: Crawlora-org (https://skillmd.com/u/crawlora-org) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/crawlora-org/used-car-market-comparison --- # Used-car market comparison Build a traceable comparison of advertised vehicles. An asking-price sample can show how a listing compares with observed alternatives; it cannot establish a transaction value, sale speed, or the vehicle's mechanical condition. ## Setup and API contract Set `CRAWLORA_API_KEY` to your key from [crawlora.net](https://crawlora.net). Use the bundled `scripts/crawlora.sh`, which sends `x-api-key` to `https://api.crawlora.net/api/v1`. Read [reference/endpoints.md](reference/endpoints.md) for selected endpoints and parameters. Successful results are inside `data`; check application `code` as well as HTTP status. Stop on authentication errors, back off on `429`, and retry a transient upstream failure once. A failed fetch is unavailable evidence, not zero inventory. Keep requests within the user's scope and credit budget; save query, page, source IDs, and retrieval times with results if collection needs to resume. ## Collect comparable listings 1. Fix the target vehicle or buyer requirements: market/ZIP, travel radius, year range, make/model, trim, powertrain, drivetrain, mileage band, condition, seller type, and budget. Keep hard requirements separate from preferences. When widening a sparse sample, show which constraint changed. 2. Search CarMax and Autotrader with supported filters. CarMax ZIP biases toward a nearby store; it is not a radius filter. Autotrader supports radius and make/model/trim codes, with model requiring make and trim requiring both. Do not reuse codes across sites; use known source values or free-text `query` and filter returned records when a code is unverified. 3. Cars.com search supports ZIP, radius, `stock_type`, and page only. Retrieve a bounded local sample and filter year/make/model/trim locally; do not invent server-side filters or present its unfiltered total as the comparable count. Keep used and certified samples distinct (`used`/`certified` on Autotrader, `used`/`cpo` on Cars.com). CarMax uses its own retail inventory context. 4. Follow returned stock numbers/listing IDs into detail for finalists and uncertain matches. CarMax detail accepts `store_id`; discover a store with the included store endpoints when local transfer context matters. Omission uses a default store and can change displayed transfer/pricing context. 5. Record source, URL, ID, VIN when returned, seller/store, actual vehicle location, price, mileage/unit, trim, condition evidence, and collection time. CarMax pages contain up to 48 results; Autotrader and Cars.com up to 24. Deduplicate within and across pages; stop repeated pages. Report budget- or pagination-limited coverage rather than claiming a complete local market. ```sh scripts/crawlora.sh /carmax/search make=Toyota model=RAV4 zip=94103 min_year=2021 max_year=2023 page=1 scripts/crawlora.sh /autotrader/search query="Toyota RAV4" condition=used zip=94103 radius=50 page=1 scripts/crawlora.sh /carsdotcom/search zip=94103 radius=50 stock_type=used page=1 ``` ## Match and compare - A verified identical VIN represents one vehicle even if advertised on multiple sites. Preserve conflicting prices and timestamps for that vehicle. Without VIN, use a provisional match supported by seller, specs, mileage, photos, and listing links; matching year/make/model alone does not establish identity. - Search and detail can label the same model differently (for example RAV4 versus RAV4 Hybrid). Retain both source labels and reconcile with VIN and powertrain detail before grouping; do not merge hybrid and gas variants by base name. - Compare exact trim, drivetrain, engine/fuel type, model year, mileage band, and certification/history context where available. Put uncertain trims, accident history, and missing mileage in separate or unknown groups. A clean-looking description is not a clean vehicle-history report. - Use numeric advertised vehicle prices in the same currency. Exclude monthly payments, down payments, missing prices, and conditional finance offers from cash-price statistics. Keep taxes, registration, dealer and transfer fees separate unless their inclusion is explicitly documented. Never label a sticker-price comparison an out-the-door quote. - Report the deduplicated comparable count, range, median, and, for a sufficiently useful sample, quartiles. A premium/discount can be `(target / median - 1) * 100` with the chosen cohort shown. Do not invent a dollars-per-mile adjustment or causal valuation model. Small or poorly matched samples warrant a descriptive range, not a precise fair-value claim. - Keep Cars.com's predicted fair price/deal rating attributed to Cars.com, separate from your sample calculations. Price history is source-reported listing history, not a transaction history. Disappearance does not prove sale. ## Deliverable Return the match criteria, coverage/stopping ledger, comparable vehicle table, price summary, and a shortlist or outlier explanation suited to the user. Include URLs, retrieval dates, excluded records/reasons, uncertain matches, fee/history gaps, and questions for the seller. Do not contact sellers, reserve vehicles, or purchase history reports unless the user requests those actions.