BeyondBSR - BookSearch API

--- name: booksearch-api description: Search Amazon KDP books on the BeyondBSR public API, retrieve the BSR (Best Sellers Rank) history of a single book, and explore the Amazon category taxonomy (browse nodes). Only the US marketplace is active; FR is temporarily disabled. Book-search results include full root-to-leaf category chains, so reports can be grouped by macro category or niche without follow-up calls. Use when the user wants to find or filter self-published or traditionally published books by BSR, category, keyword, royalty, rating, reviews, publication date or binding; do KDP niche research, low-competition discovery, sales or royalty estimates, or competitor analysis; chart the BSR timeline of an ASIN; resolve a category name into a `catId` for `categoryIds`, walk the category tree, or search categories by name; or produce category-grouped market-opportunity summaries. Do not use for price-history timelines, review or rating timelines, or account data. metadata: clawdbot: requires: env: - BOOKSEARCH_API_KEY --- # BookSearch API ## ⚠️ API Access & Beta Program The BeyondBSR BookSearch API is currently in **private beta**. This skill requires an API key (`BOOKSEARCH_API_KEY`) which is **not publicly available** at this time. Users interested in accessing Amazon KDP book data (BSR history, reviews, categories, keyword research) can apply to the early adopter program by contacting **support@beyondbsr.com**. Requests are reviewed individually and approved keys are issued on a case-by-case basis. Without a valid key, all endpoints below will return `401 Unauthorized`. --- Programmatic search over the BeyondBSR book catalogue, BSR history retrieval for a single book, and Amazon category taxonomy browsing. Six endpoints, JSON in / JSON out, API-key auth. ## When to use this skill Use it when the user asks to: - Find books matching numeric filters: BSR range, rating, reviews count, royalty, months since publication. - Discover niches by keyword inclusion/exclusion or Amazon category IDs. - Filter by marketplace (currently US / Amazon.com only; FR / Amazon.fr is temporarily disabled). - Distinguish self-publishers from traditional publishers. - Estimate sales (daily / weekly / monthly / quarterly) and revenue per copy. - Compare BSR averages across multiple time windows (7d / 30d / 90d / 180d / 365d). - Retrieve the **BSR timeline** of a single book (by `domainId` + `asin`) over the last N days, e.g. for charting rank evolution. - **Resolve a category name into an Amazon `catId`** (browse node ID), look up a category's direct children, walk its breadcrumb up to the root, or list top-level book categories for a marketplace — to feed `categoryIds` into book search, or just to explore the taxonomy. **Do NOT use** for: price-history charts, review/rating timelines, account/user data, or anything not in the response schemas below. Those are out of scope. ## Endpoints ``` POST https://api.beyondbsr.com/api/v1/books/search GET https://api.beyondbsr.com/api/v1/books/{domainId}/{asin}/bsr-history?days={1..365} GET https://api.beyondbsr.com/api/v1/categories?domainId={..}&depth={0..5}&includeFiction={true|false} GET https://api.beyondbsr.com/api/v1/categories/children?domainId={..}&catId={..} GET https://api.beyondbsr.com/api/v1/categories/search?domainId={..}&q={..}&limit={1..200} GET https://api.beyondbsr.com/api/v1/categories/ancestors?domainId={..}&catId={..} Content-Type: application/json (book search only) X-API-Key: $BOOKSEARCH_API_KEY ``` ## Authentication - Read the key from the `BOOKSEARCH_API_KEY` env var. Format: `bbsr_live_<43-char-base64url>`. - **Never** print, echo, log, or include the key in any user-facing output. Never paste it into another tool's input. - On `401 Unauthorized`: do not retry. Report "API key missing or invalid — check `BOOKSEARCH_API_KEY` env var" and stop. - Send `X-API-Key` exactly once. Multi-valued headers are rejected. ## Marketplace domains **At this time only one marketplace is active: the United States (`domainId=1`).** France (`domainId=4`) is temporarily disabled: its data is no longer refreshed, so do not use it for analysis. Additional marketplaces are planned but not yet available. Map natural-language marketplace references (e.g. "the US store", "Amazon.com") to `domainId` using this table: | domainId | locale | country | name | |----------|--------|---------|---------------| | 1 | com | US | United States | The validator technically accepts `domainId` `1–12`, but **only `1` (US) is supported right now**: other IDs may still return old rows, which must not be treated as current. If the user asks for France, tell them it is temporarily disabled. If they ask for any other marketplace (UK, Germany, Italy, Spain, Canada, Japan, etc.), tell them it is not currently supported — coming soon. Do not invent domain IDs. ## Request schema All fields are optional except `domainId`. Enums accept either the string name (e.g. `"Weekly"`) or the integer value. ### Required | Field | Type | Constraint | |------------|------|-----------------------------| | `domainId` | int | US (`1`) — the only active marketplace (validator allows 1–12). | ### BSR filters | Field | Type | Default | Notes | |------------|-------------|-----------|-----------------------------------------------------------------------------| | `bsrType` | enum | `Weekly` | `Historical(-1)`, `Weekly(7)`, `Days30(30)`, `Days90(90)`, `Days180(180)`, `Days365(365)`. `Current(0)` is rejected with `400`: read `bsrCurrent` from the response instead. | | `bsrMin` | int? | 1 | ≥ 1 | | `bsrMax` | int? | 100000 | ≥ 1, `bsrMin ≤ bsrMax` | | `bsrYear` | short? | null | 2000–(current year + 1). Required together with `bsrMonth`. Use only with `bsrType=Historical`, which returns `400` without both fields. | | `bsrMonth` | short? | null | 1–12. Required together with `bsrYear`. | ### Product filters | Field | Type | Default | Notes | |--------------------|---------|-------------|----------------------------------------------------| | `bindingType` | enum? | null (all) | `All`, `Paperback`, `Hardcover` | | `publisherType` | enum? | `All` | `All`, `SelfPublishersOnly`, `PublishersOnly`. `SelfPublishersOnly` matches only the publisher names "Independently published" and "CreateSpace", so self-publishers using their own imprint count as `PublishersOnly`. See "Aggregating results by category" before using it in niche research. | | `interiorType` | enum? | `BlackWhite`| `BlackWhite`, `FullColor` | | `vatType` | enum? | `Reduced` | `Reduced`, `Standard` | | `includePreOrders` | bool? | `false` | — | ### Quality filters | Field | Type | Constraint | |---------------|---------|------------------------------------| | `ratingMin` | double? | 0.0–5.0, `ratingMin ≤ ratingMax` | | `ratingMax` | double? | 0.0–5.0 | | `reviewsMin` | int? | ≥ 0, `reviewsMin ≤ reviewsMax` | | `reviewsMax` | int? | ≥ 0 | ### Economic filters | Field | Type | Notes | |------------------------------|----------|------------------------------------------------------| | `royaltyMin` | decimal? | In currency units, **not** cents. `min ≤ max`. | | `royaltyMax` | decimal? | — | | `monthsSincePublicationMin` | int? | `min ≤ max` | | `monthsSincePublicationMax` | int? | — | ### Discovery | Field | Type | Default | Constraint | |-------------------|----------|---------|-------------------------------------------------------------| | `includeKeywords` | string[] | null | ≤ 25 items, each ≤ 100 chars, non-empty. **Each element is matched as a literal contiguous substring** (case-insensitive) against title, publisher and authors. Multiple elements are combined with **OR**. Pass multi-word phrases as a single element (`["small business taxes"]`), NOT as separate tokens (`["small","business","taxes"]`) — the latter would match any book containing just one of those words. | | `excludeKeywords` | string[] | null | ≤ 25 items, each ≤ 100 chars, non-empty. Same matching semantics as `includeKeywords`: each element is a literal contiguous substring; books matching **any** element on title, publisher or authors are excluded. | | `categoryIds` | long[] | null | ≤ 100 items, each > 0 (Amazon BrowseNode IDs). **Subtree-expanded**: passing a non-leaf node (e.g. depth-3 "Quick & Easy", `catId=4257`) returns every book tagged with any descendant leaf — you do not need to enumerate leaves yourself. Pass any node from `/categories`, `/categories/search`, `/categories/children`, or `/categories/ancestors`. Mixing leaves and parents in one call is allowed (logical OR across the union of the expanded sets). Stale / unknown `catId`s collapse to no overlap and are silently dropped, not an error. | | `excludeFiction` | bool? | `true` | When `true` (default), filters out fiction books — i.e. books whose categories all roll up to depth-2 ancestors flagged as fiction (Literature & Fiction, Romance, Mystery, Sci-Fi, Children's Books, Teen/YA, etc.) under the local Books root for the requested marketplace. A book is kept if **at least one** of its categories has a depth-2 ancestor flagged as non-fiction. Set `false` to include fiction. **Auto-disabled** when `categoryIds` is non-empty: explicit category intent overrides the fiction filter, otherwise passing a fiction category with the default would silently return zero results. | ### Pagination | Field | Type | Default | Range | |----------|------|---------|------------| | `limit` | int? | 100 | 1–300 | | `offset` | int? | 0 | 0–100000 | ## Workflow 1. **Marketplace** — Identify `domainId` from intent using the marketplace table. If ambiguous (e.g. "Amazon"), ask which country. 2. **Translate** — Map every natural-language filter to its schema field. Don't guess enum values; use the table. Note: by default fiction is excluded (`excludeFiction=true`). If the user asks for fiction (e.g. "romance", "mystery novels", "show me fiction too") or a mixed catalog, set `excludeFiction: false` explicitly. Most KDP/low-content niche queries are non-fiction so the default is usually correct. 3. **Pagination** — Default to `limit: 50`. Raise to 100–300 only if the user explicitly wants many results. Start with `offset: 0`. 4. **Send** — POST the JSON body. Inspect the status code first. 5. **Paginate if needed** — If `returnedCount == limit`, more results likely exist. Offer to fetch the next page with `offset += limit`. 6. **Present** — Surface `title`, `asin`, `bsr`, sales/revenue estimates, and a clickable cover URL (see response notes). ## Example request ```json { "domainId": 1, "bsrType": "Days90", "bsrMin": 1, "bsrMax": 50000, "bindingType": "Paperback", "publisherType": "SelfPublishersOnly", "ratingMin": 4.0, "reviewsMin": 10, "interiorType": "BlackWhite", "vatType": "Reduced", "royaltyMin": 2.50, "royaltyMax": 15.00, "monthsSincePublicationMax": 24, "includePreOrders": false, "includeKeywords": ["journal", "notebook"], "excludeKeywords": ["coloring"], "categoryIds": [16926923011, 4742], "excludeFiction": false, "limit": 50, "offset": 0 } ``` ## Example cURL ```bash curl -X POST "https://api.beyondbsr.com/api/v1/books/search" \ -H "Content-Type: application/json" \ -H "X-API-Key: $BOOKSEARCH_API_KEY" \ -d '{ "domainId": 1, "bsrType": "Days90", "bsrMax": 50000, "bindingType": "Paperback", "publisherType": "SelfPublishersOnly", "ratingMin": 4.0, "limit": 50 }' ``` ## Response schema ### Envelope — `BookSearchApiResponse` | Field | Type | Notes | |-----------------|---------|--------------------------------------------------------------------| | `returnedCount` | int | Items in **this page**. NOT a total-match count (no total exposed).| | `limit` | int | Effective limit applied. Values outside 1–300 return `400`. | | `offset` | int | Effective offset applied. | | `results` | array | `BookSearchResultDto[]`. Empty if no match. | ### Result item — `BookSearchResultDto` (most relevant fields) | Field | Type | Notes | |----------------------|----------|----------------------------------------------------------------------------------------| | `id` | long | Internal book ID. | | `asin` | string | 10-char Amazon ASIN. | | `title` | string | Full title. | | `authors` | string? | Comma-separated, ordered by `display_order`. | | `publicationDate` | datetime?| Nullable. | | `publisher` | string? | Manufacturer/publisher name. | | `coverImageFilename` | string? | Build URL: `https://m.media-amazon.com/images/I/{filename}`. | | `imageFilenames` | string[] | All carousel images (same URL pattern). | | `rating` | double | 0.0–5.0. | | `reviews` | int | Total reviews. | | `bsr` | int | **The column matching the requested `bsrType`** — i.e. the value filtered/sorted on. | | `bsrCurrent` | int? | Latest snapshot BSR, regardless of `bsrType`. | | `avgBsr7d` | int? | Average BSR over the last 7 days. `null` if the book has no BSR data in that window. | | `avgBsr30d` | int? | Same, over 30 days. | | `avgBsr90d` | int? | Same, over 90 days. | | `avgBsr180d` | int? | Same, over 180 days. | | `avgBsr365d` | int? | Same, over 365 days. | | `pageCount` | int? | — | | `priceCents` | int? | List price (MSRP) in cents. | | `dailyEstimate` | decimal | Estimated copies/day from BSR model. | | `weeklyEstimate` | decimal | Estimated copies/week. | | `monthlyEstimate` | decimal | Estimated copies/month. | | `quarterlyEstimate` | decimal | Copies/quarter (needs ≥ 13 weeks of data). | | `royalty*Cents` | int? | Per-copy royalty in cents. 4 combinations: B/W or Color × Reduced or Standard VAT. | | `*Revenue*Cents` | long? | Derived = estimate × royalty. 16 fields total: {daily,weekly,monthly,quarterly} × {Black,Color} × {Reduced,Standard}. | | `binding` | string? | Localised label (e.g. `"Paperback"`, `"Hardcover"`, `"Non-standard"`). | | `dimensions` | string? | Formatted, prefixed by binding. Example: `"Paperback: 152 x 8 x 229 mm"`. | | `trim/spineWidthMm` | int? | Trim/spine measurements. | | `categories` | long[] | Raw leaf Amazon `cat_id`s the book is tagged with (verbatim from the database). Preserves original tag order. May be empty for books still being enriched. | | `categoryPaths` | `Array<Array<{catId:long,name:string,depth:int}>>` | One inner array per entry in `categories`, each is the full root → leaf ancestor chain sorted by `depth` ascending. Use this to aggregate / group books by macro or sub category in a single round-trip without calling `/categories/ancestors` per leaf. | | `subcategoryRanks` | `Array<{catId:long,rank:int,name:string}>` | Per-category Best Sellers Rank of the book inside each subcategory it is listed in (Amazon's "#1 in <category>" data). `catId` is the Amazon browse node, `rank` is the position within that node, `name` is the localised category label. Ordered by `rank` ascending (strongest subcategory first); the root category is excluded. May be empty `[]` for books not yet enriched. Distinct from the top-level `bsr`/`bsrCurrent`, which are the overall Books-store rank. **Do not use it to group books into niches**: books of the same niche are scattered across many sibling nodes. Use `categoryPaths` instead. | | `frequentlyBoughtTogether` | `string[]` | ASINs of products Amazon surfaces as frequently bought together with this book. Use for competitor / cross-sell discovery: feed each ASIN into `bsr-history`. `POST /books/search` has no ASIN filter, and keywords match only title, publisher and authors. May be empty `[]`. Not every book has this data — populated for a subset of titles. | **Important caveats** - All royalty and revenue fields are in **cents** — divide by 100 before showing currency. - `bsr` ≠ `bsrCurrent`. `bsr` is whatever column was chosen by `bsrType`; `bsrCurrent` is always the latest snapshot. - `avgBsrXd` is `null` when the book has no BSR data in that window, typically because it left the tracked catalogue. Do not read `null` as a bad rank or as zero, and do not assume a value exists: books with only older data may have `avgBsr365d` but not `avgBsr7d`. - In `bsrType=Historical` mode, the `avgBsrXd` averages come from the current snapshot table while `bsr` itself is the historical month value — there is a documented temporal asymmetry inside the same response. Mention this if the user is doing a strict historical analysis. - `categoryPaths.length === categories.length` for a healthy dataset. If a leaf `cat_id` has been deleted upstream after the book was indexed, the corresponding inner array is silently dropped (never null) — so `categoryPaths.length` may be ≤ `categories.length`. The `categories` array always reflects the original tag set. - Both `categories` and `categoryPaths` may be **empty arrays `[]`** (never `null`) for books still being enriched or whose taxonomy snapshot hasn't propagated yet. Aggregation code must skip these books rather than fail on missing chains. ### Aggregating results by category Each `categoryPaths` inner array is sorted by `depth` ascending. The taxonomy under the Books root is consistent across marketplaces (only the labels are localised): | Depth | Role | US example | IT example | |-------|------|------------|-----------| | 0 | Books root | `Books` | `Libri` | | 1 | Container shell | `Subjects` | `Categorie` | | 2 | **Top-level subject** ("macro") | `Self-Help`, `Cookbooks, Food & Wine`, `Crafts, Hobbies & Home` | `Cucina, casa e giardinaggio` | | 3 | **Sub-category** (usually too broad for a niche) | `Crafts & Hobbies`, `Christian Books & Bibles`, `Antiques & Collectibles` | — | | 4 | **Niche** (default grouping level, KDP-relevant) | `Coloring Books for Grown-Ups`, `Bible Study & Reference`, `Word Search`, `Coins & Medals` | — | | 5+ | **Micro-niche / leaf** (usually too fragmented to group on) | `World Coins`, `General` | — | To produce a market-opportunity report, group on the **`catId` at the desired depth**, not the `name` (names are locale-dependent; `catId` is stable). For KDP niche research, **group on depth 4 by default and fall back to depth 3 only when a chain stops there**. Example from the US tree: `Books > Subjects > Crafts, Hobbies & Home > Antiques & Collectibles > Coins & Medals > General`. - Depth 3 is too broad: `Antiques & Collectibles` has about 214k products, so coin books disappear among autographs, jewelry and stamps. - Depth 4 is the niche: `Coins & Medals` has about 9.6k products and holds all the coin books in one group. - Depth 5 and `subcategoryRanks` are too fragmented: the same books split across many nodes, and each one can fall below a minimum-books threshold. Rules for niche reports: - `General` also exists as a node name at depth 4 (e.g. `Antiques & Collectibles > General`, about 66k products). It is a catch-all, not a niche, so always show niche names with their parent, as the pattern below does. - Do not use `publisherType=SelfPublishersOnly` to decide whether a niche exists: it drops self-publishers with their own imprint and can push a real niche below a minimum-books threshold. Search with `publisherType=All`, group first, then classify publishers per niche from the `publisher` field. - Depth 4 is a default, not a guarantee: check the result when a niche looks too broad or too thin. Pattern — nested macro → sub-niche histogram with revenue rollup, from one search response: ```js // macro (depth=2) -> { name, books, subniches: Map<catId, { name, books, monthlyRevCents }> } const report = new Map(); for (const book of response.results) { const seenMacros = new Set(); // dedupe within the same book const seenNiches = new Set(); for (const chain of book.categoryPaths) { // may be [] for un-enriched books const macro = chain.find(n => n.depth === 2); const niche = chain.find(n => n.depth === 4) ?? chain.find(n => n.depth === 3); const parent = niche && chain.find(n => n.depth === niche.depth - 1); if (!macro) continue; if (!seenMacros.has(macro.catId)) { seenMacros.add(macro.catId); const m = report.get(macro.catId) ?? { name: macro.name, books: 0, subniches: new Map() }; m.books++; report.set(macro.catId, m); } if (niche && !seenNiches.has(niche.catId)) { seenNiches.add(niche.catId); const m = report.get(macro.catId); const n = m.subniches.get(niche.catId) ?? { name: `${parent.name} > ${niche.name}`, books: 0, monthlyRevCents: 0 }; n.books++; // pick whichever revenue field matches the user's interior/VAT context n.monthlyRevCents += book.monthlyRevenueBlackReducedVatCents ?? 0; m.subniches.set(niche.catId, n); } } } ``` Sample fragment of a result with two categories (US, depth-1 is always `Subjects` — depth-2 is the macro): ```json { "asin": "1234567890", "title": "Quick Weeknight Dinners", "categories": [4259, 4613], "categoryPaths": [ [ { "catId": 283155, "name": "Books", "depth": 0 }, { "catId": 1000, "name": "Subjects", "depth": 1 }, { "catId": 6, "name": "Cookbooks, Food & Wine","depth": 2 }, { "catId": 4257, "name": "Quick & Easy", "depth": 3 }, { "catId": 4259, "name": "General", "depth": 4 } ], [ { "catId": 283155, "name": "Books", "depth": 0 }, { "catId": 1000, "name": "Subjects", "depth": 1 }, { "catId": 10, "name": "Health, Fitness & Dieting", "depth": 2 }, { "catId": 4613, "name": "Diets & Weight Loss", "depth": 3 } ] ] } ``` **One-shot aggregation tip.** When the goal is a single histogram/report (as opposed to interactive paging), call `POST /books/search` with `limit=300` (the max). A 300-result response is ~850 KB and returns in ~200-350 ms in production — almost always cheaper than paginating. Only fall back to paged fetches if more than 300 books matter for the report (rare for niche analysis). ## BSR History endpoint Single-ASIN BSR timeline. Returns raw snapshots from the time-series store, ordered ascending by `recordedAt`. ``` GET https://api.beyondbsr.com/api/v1/books/{domainId}/{asin}/bsr-history?days={1..365} X-API-Key: $BOOKSEARCH_API_KEY Accept: application/json ``` ### When to use - The user has a specific ASIN and wants its BSR over time (chart, drill-down, sanity-check the search-time `avgBsrXd` averages). - The user wants to verify a book's recent rank trajectory (e.g. "is it gaining or losing visibility?"). If the user has filter criteria but no specific ASIN, use `POST /search` first, then call this endpoint per ASIN of interest. ### Path & query parameters | Param | In | Type | Required | Constraint | |------------|-------|--------|----------|----------------------------------------------------------------------------| | `domainId` | path | int | yes | US (`1`) — see Marketplace domains table (validator allows 1–12). | | `asin` | path | string | yes | Exactly 10 chars, regex `^[A-Z0-9]{10}$` (uppercase letters / digits only). | | `days` | query | int? | no | 1–365. Default `365`. Window is `[now - days, now]` UTC. | ### Example request ``` GET /api/v1/books/1/1635864348/bsr-history?days=90 X-API-Key: $BOOKSEARCH_API_KEY Accept: application/json ``` ### Example cURL ```bash curl -X GET "https://api.beyondbsr.com/api/v1/books/1/1635864348/bsr-history?days=90" \ -H "X-API-Key: $BOOKSEARCH_API_KEY" \ -H "Accept: application/json" ``` ### Response schema — `BookBsrHistoryApiResponse` | Field | Type | Notes | |---------------|-------------|--------------------------------------------------------------------| | `asin` | string | Echoed from request. | | `domainId` | int | Echoed from request. | | `fromUtc` | datetime | Window start (`now - days`), UTC. | | `toUtc` | datetime | Window end (`now`), UTC. | | `pointCount` | int | Number of BSR snapshots in `points`. | | `points` | array | `BsrPointDto[]` ordered ascending by `recordedAt`. | `BsrPointDto`: | Field | Type | Notes | |--------------------|-----------|----------------------------------------------------------------| | `recordedAt` | datetime | UTC timestamp of the snapshot. | | `bestSellersRank` | int? | BSR at that timestamp. May be null if Keepa returned no rank. | Example body: ```json { "asin": "1635864348", "domainId": 1, "fromUtc": "2026-01-27T00:00:00Z", "toUtc": "2026-04-27T00:00:00Z", "pointCount": 412, "points": [ { "recordedAt": "2026-01-27T03:14:00Z", "bestSellersRank": 1234 }, { "recordedAt": "2026-01-27T15:02:00Z", "bestSellersRank": 1218 } ] } ``` ### Status codes (BSR history specific) | Code | Meaning | Notes | |------|-------------------------------|-----------------------------------------------------------------------| | 200 | OK | `pointCount` may be 0 if no snapshots exist in the window. | | 400 | Validation failed | `ValidationProblemDetails`. Common causes: `days` out of range, `asin` wrong format, `domainId` out of `[1,12]`. | | 404 | ASIN not found for that domain| `ProblemDetails` JSON (`{title:"Not Found",status:404,...}`). The book is not tracked in BeyondBSR for this marketplace. Tell the user — do not retry with the same pair. | | 401 / 429 / 500 | See generic table below. | ### Caveats - Granularity is **raw**: Keepa snapshots arrive 1–4× per day on actively-tracked books. A 365-day window typically yields 365–1460 points. No daily aggregation is applied. - The endpoint returns only `(recordedAt, bestSellersRank)`. Use `POST /search` if you also need rating/review/price data. - ASIN is case-sensitive — must be uppercase. `1635864348` (all digits) is valid. - Default 365 days is the cap. Longer histories are not exposed; do not retry with `days > 365`. ## Categories endpoints The four endpoints under `/api/v1/categories` expose the Amazon taxonomy (browse nodes) BeyondBSR has ingested per marketplace. They share the same API-key auth, rate limit (`api-key` policy), and `domainId` semantics as book search. All four are `GET`, JSON out, no request body. ### Use cases - **Resolve a name → `catId`** to pass into `POST /books/search` `categoryIds`. Example: user asks for "manga" books on US → call `/categories/search?domainId=1&q=manga` to get candidate cat_ids, then feed the chosen ones into `categoryIds`. - **Top-level book categories** of a marketplace (e.g. "list the main Books categories on Amazon.com"): `GET /categories?domainId=1&depth=2`. - **Walk down the tree** from a known node (e.g. "what's under Cookbooks?"): `GET /categories/children?domainId=1&catId=6`. - **Walk up the tree / breadcrumb** for a leaf node (e.g. "where does cat 11160395011 sit?"): `GET /categories/ancestors?domainId=1&catId=11160395011` → returns root → … → node, ordered by depth. ### Shared response object — `CategoryBrowseNodeDto` | Field | Type | Notes | |-------------------|----------|---------------------------------------------------------------------------------------------| | `catId` | long | Amazon browse node ID. **This is the value to pass into `categoryIds` in book search.** | | `parentCatId` | long? | Parent's `catId`. `null` for roots. | | `rootCatId` | long | Top-level ancestor's `catId` (e.g. the Books root for the marketplace). | | `name` | string | Localised name in the marketplace's language. | | `contextFreeName` | string? | Name without parent context (Keepa-provided). May be null. | | `depth` | int | 0 = root. Top-level book categories are typically depth 2 (root → Categorie → top node). | | `isFiction` | bool | Internal flag used by `excludeFiction` in book search. Only meaningful at depth=2 under Books root. | | `productCount` | int? | Approximate number of products in that node (Keepa-reported). May be null. | ### 1. List categories at a depth (top-level Books browser) ``` GET /api/v1/categories?domainId={..}&depth={0..5}&includeFiction={true|false} ``` | Param | Type | Default | Notes | |------------------|-------|---------|----------------------------------------------------------------------------------------------------------------------------------------| | `domainId` | int | — | Required. Marketplace ID (1-12, but only `1` US is supported). | | `depth` | int? | 2 | Hierarchy level. 0 = root. 2 = top-level book categories ("Self-Help", "Cookbooks…", etc.). Allowed 0-5. | | `includeFiction` | bool? | `false` | Whether to include nodes flagged as fiction (Literature & Fiction, Romance, Sci-Fi, Children's Books, Teen/YA, Comics, etc.). | **Filters applied automatically (do not appear in params):** only browse nodes (`isBrowseNode=true`), and only nodes whose root is the **Books** root of the marketplace — i.e. non-book trees (toys, electronics) are excluded. Use this endpoint to enumerate the canonical KDP-relevant taxonomy. Order: `productCount` DESC, then `name` ASC. Response envelope — `CategoriesApiResponse`: | Field | Type | Notes | |-----------------|---------|--------------------------------------| | `domainId` | int | Echo. | | `depth` | int | Echo (resolved default if omitted). | | `returnedCount` | int | Number of items in `results`. | | `results` | array | `CategoryBrowseNodeDto[]`. | ### 2. Direct children of a category ``` GET /api/v1/categories/children?domainId={..}&catId={..} ``` | Param | Type | Notes | |------------|------|----------------------------------------------------------------------------------| | `domainId` | int | Required. | | `catId` | long | Required. Parent category's Amazon browse node ID. | **No implicit filters.** Returns ALL active direct children of the parent — including non-browse nodes and fiction nodes. Use it for true tree navigation regardless of book/non-book context. Order: `productCount` DESC, then `name` ASC. Returns **404** if `catId` does not exist for the given marketplace. Do not retry on 404. Response envelope — `CategoryChildrenApiResponse`: | Field | Type | Notes | |-----------------|--------|--------------------------------------| | `domainId` | int | Echo. | | `parentCatId` | long | Echo of the input `catId`. | | `returnedCount` | int | | | `results` | array | `CategoryBrowseNodeDto[]`. | ### 3. Search categories by name ``` GET /api/v1/categories/search?domainId={..}&q={..}&limit={1..200} ``` | Param | Type | Default | Notes | |------------|---------|---------|--------------------------------------------------------------------------------------------------------------------| | `domainId` | int | — | Required. | | `q` | string | — | Required. Case-insensitive substring match on `name`. Min **3** chars, max 100. Wildcards (`%`, `_`, `\`) are treated as literals (escaped server-side). | | `limit` | int? | 50 | 1-200. Pushed down to SQL. | **No implicit filters.** Searches across the ENTIRE category tree of the marketplace (not just Books) — so a query like "Sports" on Amazon.com will surface both "Sports & Outdoors" (under Books) and the toy/apparel "Sports" nodes. The agent should filter client-side by `rootCatId` if the user only wants book categories. Order: `productCount` DESC, then `depth` ASC, then `name` ASC. Response envelope — `CategorySearchApiResponse`: | Field | Type | Notes | |-----------------|--------|--------------------------------------| | `domainId` | int | Echo. | | `query` | string | Echo of `q`. | | `limit` | int | Effective limit applied. | | `returnedCount` | int | | | `results` | array | `CategoryBrowseNodeDto[]`. | **Tip:** if the user-typed term is broad (e.g. "fiction") and the default `limit=50` truncates likely candidates, raise `limit` to 200. If still not enough, refine the term or use `/categories?depth=2` instead. ### 4. Ancestors / breadcrumb of a category ``` GET /api/v1/categories/ancestors?domainId={..}&catId={..} ``` | Param | Type | Notes | |------------|------|--------------------------------------------------------| | `domainId` | int | Required. | | `catId` | long | Required. Category whose breadcrumb to retrieve. | Returns the full ancestor chain **including the node itself**, ordered by `depth` ASC. No implicit filters (intermediate non-browse / promotional nodes are included). Returns **404** if `catId` does not exist for the marketplace. Response envelope — `CategoryAncestorsApiResponse`: | Field | Type | Notes | |-----------------|--------|----------------------------------------------------------------------| | `domainId` | int | Echo. | | `catId` | long | Echo. | | `returnedCount` | int | Length of the chain (includes the requested node). | | `results` | array | `CategoryBrowseNodeDto[]` from root (depth=0) to the node itself. | ### Example workflows **Resolve "manga" on Amazon.com then search books** (replace `<catId>` with a candidate returned by step 1): ```bash # 1. find candidate cat_ids curl -H "X-API-Key: $BOOKSEARCH_API_KEY" \ "https://api.beyondbsr.com/api/v1/categories/search?domainId=1&q=manga&limit=10" # 2. inspect the breadcrumb of a candidate to confirm it's under Books curl -H "X-API-Key: $BOOKSEARCH_API_KEY" \ "https://api.beyondbsr.com/api/v1/categories/ancestors?domainId=1&catId=<catId>" # 3. drill down children if needed curl -H "X-API-Key: $BOOKSEARCH_API_KEY" \ "https://api.beyondbsr.com/api/v1/categories/children?domainId=1&catId=<catId>" # 4. plug the chosen cat_id(s) into book search curl -X POST -H "X-API-Key: $BOOKSEARCH_API_KEY" -H "Content-Type: application/json" \ -d '{"domainId":1,"categoryIds":[<catId>],"limit":50}' \ "https://api.beyondbsr.com/api/v1/books/search" ``` **List top-level Books categories on Amazon.com (non-fiction only, default):** ```bash curl -H "X-API-Key: $BOOKSEARCH_API_KEY" \ "https://api.beyondbsr.com/api/v1/categories?domainId=1&depth=2" ``` ### Status codes (categories endpoints) | Code | Meaning | Notes | |------|----------------------------------|---------------------------------------------------------------------------------------------| | 200 | OK | `results` may be empty (e.g. no children, no matches, marketplace not seeded). | | 400 | Validation failed | `ValidationProblemDetails`. Common causes: `domainId` out of range, `q` < 3 chars, `limit` out of 1-200, `depth` out of 0-5. | | 401 | Auth failed | Same handling as book search — stop, report misconfigured key. | | 404 | `catId` not found for `domainId` | `ProblemDetails` JSON (`{title:"Not Found",status:404,...}`). Only on `/children` and `/ancestors`. Do not retry the same pair. | | 429 | Rate limit | Honour `Retry-After`. Same key budget as book search (30 req/min). | | 500 | Server error | Retry once, then escalate. | ### Caveats - **Read-only.** No POST/PUT/DELETE on categories. - **`catId` vs internal id.** The wire contract exposes only Amazon's `catId` (browse node ID). The internal database `id` is never surfaced and is not interchangeable. - **`depth=2` ≠ "top-level under Books" universally.** Most marketplaces seed Books root at depth=0 → "Categorie/Categories" at depth=1 → top nodes at depth=2. If `/categories?depth=2` returns unexpectedly few rows for a marketplace, retry with `depth=1`. - **`excludeFiction` semantics propagate.** A node's `isFiction=true` here is exactly what `excludeFiction` in book search rolls up on. If a user complains that a fiction sub-genre isn't being excluded, the agent can spot-check via `/categories/ancestors?catId=…` whether the depth-2 ancestor is flagged. - **Children endpoint includes non-browse nodes.** Some Amazon nodes are promotional or grouping shells (`isBrowseNode=false`). They cannot be used as `categoryIds` filters in book search. Skip them or warn the user. ## Status codes & error handling | Code | Meaning | Body | Agent action | |------|------------------------|--------------------------------------------|-------------------------------------------------------------------------------| | 200 | OK (may be empty list) | `BookSearchApiResponse` | Parse `results`. Empty array = no matches, not an error. | | 400 | Validation failed | `ValidationProblemDetails` (RFC 7807) | Read the `errors` map, fix the body, do not retry blindly. Surface the issue. | | 401 | Auth failed | empty + `WWW-Authenticate: ApiKey realm="BeyondBSR"` | Stop. Report misconfigured `BOOKSEARCH_API_KEY`. Do not retry. | | 429 | Rate limit exceeded | `ProblemDetails` JSON (`detail`: "Rate limit exceeded. Please try again later.") + `Retry-After` header | Honour `Retry-After`. Back off. Do not hammer the endpoint. | | 500 | Unhandled server error | `ProblemDetails` JSON | Retry once after a few seconds. If it persists, escalate to the user. | ### Example 400 body ```json { "type": "https://tools.ietf.org/html/rfc7231#section-6.5.1", "title": "One or more validation errors occurred.", "status": 400, "errors": { "DomainId": ["DomainId must be between 1 and 12."], "Limit": ["Limit must be between 1 and 300."] } } ``` ## Sorting & pagination notes - Default sort: `ORDER BY <chosen BSR column> ASC` (lower BSR = better seller appears first). - Historical mode (`bsrType=Historical` + `bsrYear`/`bsrMonth`): sorted by the historical monthly BSR rank ASC. - Tie-breaker is non-deterministic — books with identical BSR may appear in different orders across page fetches. Warn the user if they need a perfectly stable ordering. - No `orderBy` parameter is exposed. ## Limits - **Rate limit**: 30 requests/minute per API key. Shared across all callers using the same key. - **Body size**: 128 KB max. - **Page size**: 300 results max per request (`limit ≤ 300`). - **Offset cap**: 100,000 (deep pagination beyond this is not supported — refine filters instead).

存档时间线

版本索引永久保留；内容副本只保留最新 1 份，更早版本仅留索引与哈希（存档时间线的证据链不会因此断裂）。

下载存档内容副本