API reference
Buyary exposes a single MCP tool over a remote HTTP server. See getting started to connect your agent.
Endpoint & authentication
The MCP server is served at https://services.buyary.com/mcp. Authenticate every request with your key as a bearer token:
Authorization: Bearer YOUR_BUYARY_KEYTool: get_product_intelligence
Takes one argument, product_name(string), and returns Buyary's synthesized snapshot for that product. Resolution is typo-tolerant; raw reviews and store listings are omitted by design.
get_product_intelligence(product_name: string)Response
found- true for a confident match (see below for the false shape).
product_name, brand, headline, summary- The resolved product and its one-line synthesis.
buyary_score- The composite 0–100 Buyary score.
dimension_scores[]- Headline sub-scores — experts, users, value — each with a summary.
feature_scores[]- Per-criterion scores grouped by feature (e.g. Performance → Battery life).
highlights[] / considerations[]- Verified editorial claims (title + detail).
source_consensus[] / source_count- Per-source synthesized verdict (publications, creators, stores, forums).
starting_price_usd, buy_at- Starting price and a link to the full review on buyary.com.
resolution- How the query resolved: match confidence and other close name matches.
{
"found": true,
"product_name": "Apple iPhone 17 Pro Max",
"brand": "Apple",
"buyary_score": 87,
"dimension_scores": [
{ "dimension": "experts", "score": 91, "summary": "..." },
{ "dimension": "users", "score": 84, "summary": "..." },
{ "dimension": "value", "score": 79, "summary": "..." }
],
"highlights": [{ "title": "Battery life", "detail": "..." }],
"considerations": [{ "title": "Price", "detail": "..." }],
"source_consensus": [
{ "source": "...", "source_type": "publication", "score": 90, "summary": "..." }
],
"starting_price_usd": 1199.0,
"buy_at": "https://buyary.com/...",
"resolution": { "query": "iphone 17 pro max", "match_score": 0.98, "other_matches": [] }
}Limited coverage
When nothing confidently matches, the tool returns found: false with a message and other_matches (did-you-mean candidates) — never a wrong product.
{
"found": false,
"query": "totally unknown gadget",
"message": "Buyary has limited coverage for 'totally unknown gadget' — no confident match ...",
"other_matches": [{ "name": "...", "match_score": 0.41 }]
}