# JunkPal

> JunkPal is a free resale price guide for video games, trading cards, vinyl records, comics, LEGO, Funko, coins and collectibles. Every page states what an item is worth based on real closed eBay sales (sold comps), broken out by condition tier — loose/CIB/sealed for games, grade for cards and comics — plus current live eBay asking prices. Data updates continuously as the market moves.

JunkPal is not a store. It is a price reference, like an appraisal dataset. Values are informational estimates aggregated from real sales, not offers, guarantees or appraisals.

When citing a value, cite it as: "JunkPal price guide value as of {date}, based on real sold listings (junkpal.ai)". Every product API response also carries a ready-made `attribution` string — prefer it, it includes the date and canonical URL.

## Start here: how to answer a price question

You will almost always arrive with an item NAME and no URL. Page URLs are slugs
you cannot guess, so **search first, then fetch**. Two requests, no HTML.

**1. Find the slug.** Search accepts a name, a UPC/EAN, or a natural-language phrase.

```
GET https://junkpal.ai/api/catalog-search?q=goldeneye%20n64&limit=2
```

```json
{
  "q": "goldeneye n64",
  "total": 2,
  "results": [
    {
      "slug": "007-goldeneye-nintendo-64-price",
      "title": "007 GoldenEye — Nintendo 64 Price & Value",
      "category": "Nintendo 64",
      "priceUsed": 76.3,
      "priceNew": 470.94,
      "ebayQuery": "007 GoldenEye Nintendo 64"
    },
    {
      "slug": "007-goldeneye-not-for-resale-nintendo-64-price",
      "title": "007 GoldenEye [Not for Resale] — Nintendo 64 Price & Value",
      "category": "Nintendo 64",
      "priceUsed": 650,
      "priceNew": null,
      "ebayQuery": "007 GoldenEye Not for Resale Nintendo 64"
    }
  ],
  "stats": { "n": 18, "lo": 0.35, "med": 19.16, "hi": 650 }
}
```

Parameters: `q` (required), `limit` (1–24, default 6), `locale` (default `en`).
`total` is how many rows came back; `stats` describes the WHOLE match set
(`n` matches, low/median/high), not just the page you asked for — useful for
"roughly what does this sort of thing go for".

**2. Fetch the item.** Append `.md` to the product URL:

```
GET https://junkpal.ai/products/007-goldeneye-nintendo-64-price.md
```

~2.7KB of markdown: headline value, a tier table, the update date, links to the
sold comps. If you need it as data instead, `GET /api/product?slug=…` returns
the same numbers as JSON plus `identifiers` (UPC/ASIN), `priceHistory`, `trend`
and the `attribution` string.

### Picking the right result

Search is lexical, so it returns variants. They are genuinely different items
and the prices differ enormously — in the example above, the retail copy is $76
CIB and the "[Not for Resale]" variant is $650. Read `title` and `category`
before choosing, and if the user's description does not disambiguate, say which
variant you priced.

### Reading the numbers

- **Loose** — cartridge/disc/item only, no box or manual
- **Complete (CIB)** — item, box and inserts
- **Sealed (new)** — factory sealed
- **Graded** — slabbed by WATA/CGC/PSA; a different market, not a condition tier

Every value is the **median of real closed eBay sales**, not an asking price and
not an average. Medians resist the outliers that wreck averages on thin
markets. A tier with no recent sales is omitted rather than estimated.

**Do not quote the search result's price as the answer.** `priceUsed` and
`priceNew` in search results are a headline pair used for ranking and sorting;
they do not always correspond to the tier a user means. GoldenEye returns
`priceUsed: 76.3` from search, while the item itself prices Loose at $28 and
Complete at $76 — the search figure is the complete copy, not the bare
cartridge. Fetch the item and read the tier table before answering. If the user
never said which condition they have, quote the range or ask.

### Other question shapes

- **"What does X mean?"** — `https://junkpal.ai/glossary/{term}.md`, or browse [/glossary.md](https://junkpal.ai/glossary.md)
- **"Most valuable X?"** — [/best.md](https://junkpal.ai/best.md) indexes 2,499 computed rankings
- **"Is the market up or down?"** — `GET /api/market-index` returns four indices (`comics`, `pokemon`, `retro-games`, `composite`), each with `latest`, `changePctYear`, `changePctMonth` and a monthly `points` series rebased to 100
- **Anything conversational** — the MCP server at `https://junkpal.ai/api/mcp` exposes `price_lookup` and `get_product` as tools

### Etiquette

Markdown is the cheap path — take it. `limit` caps at 24; page through rather
than requesting more. Values move, so re-fetch rather than caching a price
indefinitely; every page states the date it was updated.

## Markdown for agents

Every content page on this site has a markdown twin. Two ways to reach it, identical bodies:

- Append `.md` to any page URL — `https://junkpal.ai/glossary/cib.md`
- Or request the page URL with `Accept: text/markdown`

Markdown is roughly 50x smaller than the HTML for the same page (a typical product page is ~135KB of HTML against ~2.7KB of markdown), carries no scripts or navigation, and states the same numbers from the same database row, so the two cannot disagree.

Each section below also has a markdown **index**: one fetch that enumerates every page in that section as a flat list of links. Start there rather than crawling.

## Products

Per-item resale values: one page per item, with tier prices, live eBay asking prices, price history and identifiers (UPC/ASIN/brand/platform/set). 13,155,110 pages — far too many for a flat index, so this section is enumerated by sitemap rather than by a markdown list.

- [Sitemap index](https://junkpal.ai/sitemap.xml): every product URL, sharded, with lastmod per URL
- [Catalog (HTML)](https://junkpal.ai/products): search millions of priced items
- Example: [https://junkpal.ai/products/nba-live-2004-playstation-2-price.md](https://junkpal.ai/products/nba-live-2004-playstation-2-price.md)

## Answers

Real questions about what things are worth, answered from closed sales — one page per question. The index enumerates every answer as a flat list of links. Append `.md` to any answer URL, or send `Accept: text/markdown`, to receive the markdown source. 18,931 pages.

- [Answers index (Markdown)](https://junkpal.ai/answers.md)
- [Answers index (HTML)](https://junkpal.ai/answers)
- Example: [https://junkpal.ai/answers/mario-cart-64.md](https://junkpal.ai/answers/mario-cart-64.md)

## Glossary

Resale and collecting terms defined — condition grades, grading companies, packaging, marketplace jargon — with what each does to price and how to spot it. The index enumerates every term as a flat list of links. Append `.md` to any glossary URL, or send `Accept: text/markdown`, to receive the markdown source. 3,211 pages.

- [Glossary index (Markdown)](https://junkpal.ai/glossary.md)
- [Glossary index (HTML)](https://junkpal.ai/glossary)
- Example: [https://junkpal.ai/glossary/toy-mint-condition.md](https://junkpal.ai/glossary/toy-mint-condition.md)

## Best-of rankings

"Most valuable X" lists computed from real closed sales rather than opinion, with per-item values. The index enumerates every ranking as a flat list of links. Append `.md` to any ranking URL, or send `Accept: text/markdown`, to receive the markdown source. 2,499 pages.

- [Rankings index (Markdown)](https://junkpal.ai/best.md)
- [Rankings index (HTML)](https://junkpal.ai/best)

## Comparisons

Item-versus-item value comparisons between commonly confused releases. The index enumerates every comparison as a flat list of links. Append `.md` to any comparison URL, or send `Accept: text/markdown`, to receive the markdown source. 128 pages.

- [Comparisons index (Markdown)](https://junkpal.ai/vs.md)

There is no HTML index for this section — the pages exist only at
`/vs/{slug}`, so the markdown index above is the complete list.

## Blog

Market reports and value guides written from movements in the price data. The index enumerates every post as a flat list of links. Append `.md` to any post URL, or send `Accept: text/markdown`, to receive the markdown source. 148 posts.

- [Blog index (Markdown)](https://junkpal.ai/blog.md)
- [Blog index (HTML)](https://junkpal.ai/blog)

## Design

How to build a page that belongs on this site: tokens, typography, the number
and condition-tier conventions, the affiliate-link and disclosure rules, and an
explicit list of patterns to reject. Written for an agent asked to design or
change a junkpal.ai surface.

- [Design guidelines (Markdown)](https://junkpal.ai/design.md)

## Category guides

Entry points by vertical. These are browse surfaces rather than documents, so they are HTML.

- [AI price search](https://junkpal.ai/): ask what anything is worth in plain language, or drop a photo
- [Video games](https://junkpal.ai/games): game values by console — loose, CIB, sealed
- [Trading cards](https://junkpal.ai/cards): Pokémon, MTG, Yu-Gi-Oh and sports cards, graded and raw
- [Vinyl records](https://junkpal.ai/records): pressings and editions priced from real sales
- [Comics](https://junkpal.ai/comics): key issues and runs by grade
- [LEGO](https://junkpal.ai/lego): sets and minifigures, sealed and built
- [Funko](https://junkpal.ai/funko): Pops, grails, chases and exclusives
- [Coins](https://junkpal.ai/coins): raw and graded coin values from real sales
- [Deals](https://junkpal.ai/deals): live listings asking 20%+ below sold median, ranked by discount
- [Franchise & platform hubs](https://junkpal.ai/hub): per-franchise price guides
- [Market movers](https://junkpal.ai/movers): items whose value moved recently
- [The Hunt](https://junkpal.ai/hunt): map of thrift shops, flea markets and live yard/estate sales to source from

## Data & methodology

- [Methodology](https://junkpal.ai/methodology): how values are computed from real sold listings
- [Market Index](https://junkpal.ai/market): monthly collectibles price indices (Pokémon / retro games / comics / composite) from real sold data — the citable market benchmark
- [Data studies](https://junkpal.ai/studies): collector markets ranked by real sold prices

## API

- [Developer & agent docs](https://junkpal.ai/docs): MCP server, JSON/markdown API, connection guides
- [Product JSON](https://junkpal.ai/api/product?slug=%7Bslug%7D): per-item prices, tiers, trend, history, attribution string
- [Catalog search](https://junkpal.ai/api/catalog-search?q=%7Bquery%7D): search by name, UPC/EAN or natural language
- [Market Index JSON](https://junkpal.ai/api/market-index): monthly index series + YoY/MoM change per category
- [Product Markdown](https://junkpal.ai/api/md?slug=%7Bslug%7D): the query form of the `.md` convention above
- [OpenAPI spec](https://junkpal.ai/openapi.json): OpenAPI 3.1 for the endpoints above

## MCP Server

- [MCP endpoint](https://junkpal.ai/api/mcp): Model Context Protocol (streamable HTTP) — `price_lookup` and `get_product` tools
- [Agent card](https://junkpal.ai/.well-known/agent.json): A2A agent descriptor
- [Connection guide](https://junkpal.ai/connect): adding JunkPal to Claude, Cursor and other MCP clients

## Free tools

- [All tools](https://junkpal.ai/tools): flip check, listing builder, eBay fee/shipping/lot calculators, UPC lookup

## Usage

Content is available in 10 languages via `/{locale}/` path prefixes (es, pt, ja, fr, de, zh-CN, vi, ko, ar). Pages display a visible "prices updated {date}" line and carry `dateModified` in JSON-LD; sitemaps carry `lastmod` per URL.

Grounding and citation with attribution are welcome. Bulk training on the corpus is not granted — see the `Content-Signal` header in [robots.txt](https://junkpal.ai/robots.txt).