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, page count age, publication recency.
• Discover niches by keyword inclusion/exclusion or Amazon category IDs.
• Filter by marketplace (currently US / Amazon.com and FR / Amazon.fr only).
• 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 two marketplaces are populated with data: the United States (domainId=1) and France (domainId=4). Additional marketplaces are planned but not yet available.

Map natural-language marketplace references (e.g. "the US store", "Amazon.com", "amazon francia", "Amazon.fr") to domainId using this table:

The validator technically accepts domainId 1–12, but only 1 (US) and 4 (FR) return data right now. If the user asks 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

BSR filters

Product filters

Quality filters

Economic filters

Discovery

Pagination

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

{
  "domainId": 4,
  "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": [266162, 3248921],
  "excludeFiction": false,
  "limit": 50,
  "offset": 0
}

Example cURL

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": 4,
    "bsrType": "Days90",
    "bsrMax": 50000,
    "bindingType": "Paperback",
    "publisherType": "SelfPublishersOnly",
    "ratingMin": 4.0,
    "limit": 50
  }'

Response schema

Envelope — BookSearchApiResponse

Result item — BookSearchResultDto (most relevant fields)

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.
• 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):

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 the interesting depths are 3 and 4: depth 2 is usually too broad (e.g. "Crafts & Hobbies" alone covers thousands of books), while depth 3-4 isolates real nicchie ("Coloring Books for Grown-Ups", "Word Search", "Christian Living").

Pattern — nested macro → sub-niche histogram with revenue rollup, from one search response:

// 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 === 3) ?? chain.find(n => n.depth === 4);
    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: 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):

{
  "asin": "1234567890",
  "title": "Quick Weeknight Dinners",
  "categories": [4259, 9876],
  "categoryPaths": [
    [
      { "catId": 283155, "name": "Books",                "depth": 0 },
      { "catId": 1000,   "name": "Subjects",             "depth": 1 },
      { "catId": 6,      "name": "Cookbooks, Food & Wine","depth": 2 },
      { "catId": 17,     "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": 9876,   "name": "Diet & 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

Example request

GET /api/v1/books/1/1635864348/bsr-history?days=90
X-API-Key: $BOOKSEARCH_API_KEY
Accept: application/json

Example cURL

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

BsrPointDto:

Example body:

{
  "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)

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.fr"): GET /categories?domainId=4&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 4142740011 sit?"): GET /categories/ancestors?domainId=1&catId=4142740011 → returns root → … → node, ordered by depth.

Shared response object — CategoryBrowseNodeDto

1. List categories at a depth (top-level Books browser)

GET /api/v1/categories?domainId={..}&depth={0..5}&includeFiction={true|false}

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:

2. Direct children of a category

GET /api/v1/categories/children?domainId={..}&catId={..}

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:

3. Search categories by name

GET /api/v1/categories/search?domainId={..}&q={..}&limit={1..200}

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:

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={..}

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:

Example workflows

Resolve "manga" on Amazon.com then search books (replace <catId> with a candidate returned by step 1):

# 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):

curl -H "X-API-Key: $BOOKSEARCH_API_KEY" \
  "https://api.beyondbsr.com/api/v1/categories?domainId=1&depth=2"

Status codes (categories endpoints)

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

Example 400 body

{
  "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).