io.github.Swervie由 GitHub 帳號 Swervie 發布v1.6.0更新於 Oct 29, 2026
Cross-retailer product catalog: one scored answer per product with live prices and buy links.
Swervie 是跨零售商的商品目錄，讓助理用一份整合、評分的商品答案回答購物查詢，並附上即時價格與購買連結。
Web executable，透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。
把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式，或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南，先向你展示發布者、端點和登入方式，經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。
閱讀 https://sourceweft.com/mcp/io.github.Swervie%2Fswervie/install.md，並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前，先告訴我它的發布者、端點和登入方式，等我確認。不要把 API 金鑰寫進命令或我會提交的檔案；問我金鑰應從哪裡取得。把它新增到你客戶端的 mcpServers 設定中。
{
"mcpServers": {
"swervie": {
"type": "http",
"url": "https://us-central1-swervie-prod.cloudfunctions.net/mcp"
}
}
}
One scored answer per product, across every retailer, for AI agents.
Remote MCP server · Streamable HTTP · no auth · no API key · free
Developer docs · Tool reference · Connect a client · Response shapes · Data handling
Swervie is a cross-retailer product catalog built for assistants. Ask a normal shopping question and get one consolidated, scored answer per product: live prices, comparable specifications, review-derived ratings with verbatim evidence quotes, and a buy link for every seller. Instead of eight duplicate listings and a stale listicle.
io.modelcontextprotocol/ui extension); plain JSON everywhere else.Coverage today: 300+ categories, from robot vacuums and televisions to espresso machines, mattresses and hiking boots, across hundreds of sellers. The live list is at swervie.xyz/llms.txt.
Claude Code
Cursor, Windsurf, VS Code, Claude Desktop, ChatGPT, Gemini CLI and others: see docs/clients.md. Every config is the same one line, the URL above, with no headers and no secrets.
curl
More in examples/: a Node client using the official SDK, a Python client, and the curl script above.
Every tool is read-only and annotated (title, readOnlyHint: true,
openWorldHint: true), so hosts can run them without a confirmation prompt.
The server provides data only: it writes nothing on the user's behalf, and
neither its instructions nor its results direct the model. Full input schemas
and descriptions, generated from the live server, are in
docs/tools.md.
search_products → ranked shelf
with prices, buy links and the commission disclosure.get_products → refresh rate, VRR,
HDMI 2.1, and what reviewers say about input lag.lookup_product → every seller
offer, plus its rank on the shelf and higher-scoring alternatives.The server ships instructions on initialize that describe each tool and
the fields its results carry, so a client needs no prompt engineering.
Every tool accepts an optional user_signature: an opaque, stable, pseudonymous
id that gives each shopper their own rate-limit quota (instead of sharing one
per session or egress IP) and attributes their buy-link clicks. The
recommended recipe:
where id is the user's most stable identifier, normalized (lowercase,
trimmed; email preferred, with +tags stripped). Never send the raw
identifier. If none is available, use "sv1r-" + a UUID kept for that user,
or omit the field. A value that changes every session counts as a new shopper
each session. The server hashes any raw email or phone number that arrives by
mistake before anything is stored.
[email protected].buy_url
also carries a disclosure string worded for the shopper. Commissions never
affect rankings, scores, or which products are shown.Details in docs/data-handling.md. Policies: privacy · terms · disclosure.
This repository is the public home of the Swervie MCP server: documentation,
the registry manifest (server.json), client configuration, examples, and
brand assets for directories. The serving code runs as a Google Cloud Function
inside Swervie's main monorepo and is not published here. The hosted endpoint
is the supported way to use it. Issues and feature requests are welcome.
Server version is reported by initialize (serverInfo.version). Protocol
version 2025-06-18. Transport: Streamable HTTP, stateless.
Changes: v1.6.0 (2026-10) removed the two non-destructive write tools
(record_selection, submit_feedback) and rewrote the server instructions
and result notes as descriptions of the data; the server is now read-only
end to end.
Documentation, examples and assets in this repository are MIT licensed (© Plum Applications, Inc.). Use of the hosted service is governed by the terms.
來源：README.md，提交 526795e
Find a product category
Find the Swervie product category that matches a shopping intent and return its filter contract: every filterable specification with its type, units, allowed values, and importance, plus budget, mid, and premium price bands.
query必填user_signatureBrowse a category shelf
Return a category's consolidated, ranked shelf: one row per product, with duplicate seller listings already collapsed. Each row carries price, price_band, buy_url and buy_seller (the retailer behind the link), image_url, star ratings, and the category's required and top-weighted spec values. card_specs gives the specs every product card shows, in display order. Filters are hard: a row missing the attribute is excluded. Preferences are soft: rows that measure well on them rank higher and none are excluded (for example Rated Noise, prefer low, or a pet-hair rating, prefer high). Sort by value (default), quality, or price_asc. A filter on the Price dimension applies min/max to the live offer price in cents. Responses that carry buy_url links also carry a disclosure field: the commission disclosure for those links.
sortlimitcursorfilterscategory必填preferencesuser_signatureSearch products
Answer one free-text shopping query, such as 'TVs under $500 with 120hz refresh rate', 'quiet dishwasher for a small kitchen', 'robot vacuum for pet hair under $600', or 'Sony WH-1000XM5 lowest price'. A query naming one specific product resolves to that product (interpretation.mode 'product'): products[0] holds every seller offer, and shelf_context gives its shelf rank and higher-scoring alternatives. 'A vs B' returns both products with specs in one shared order (mode 'comparison'). Otherwise the category is resolved, hard constraints such as price and specs become filters, and soft wishes such as 'quiet' or 'for pet hair' become ranking preferences that exclude nothing; interpretation.preferences lists them with reasons, and each row's preference_boost shows the effect. The interpretation object records how the query was read: method (deterministic parser or query-understanding model), fit_note (a one-line judgment of the category fit), and match_quality. 'strong' means the shelf plainly fits. 'partial' means either rows are served with some shopper words unexplained (listed in unparsed_tokens and unsupported_terms), or no rows are served because the closest shelf was only a medium-confidence fit (named in interpretation.llm_category). 'weak' and 'none' return no rows and queue the request as demand (queued_for_creation for a missing category, queued_product_request for a missing product). card_specs gives the specs every product card shows, in display order. Responses that carry buy_url links also carry a disclosure field: the commission disclosure for those links.
sortlimitquery必填cursoruser_signatureLook up a specific product
Look up one specific product by name or identifier: a product name such as 'Sony WH-1000XM5' or 'Ninja AF101 air fryer', an ASIN, a UPC/EAN/GTIN, or brand plus model number. match reports the result: 'exact' (identifier hit), 'likely' (one clear name match; interpretation.runner_up names the next best), 'ambiguous' (several fit; rows lists the candidates, best first), or 'none' (not carried; the miss is recorded as demand). On exact or likely, products[0] is the full detail (all specs, review-derived ratings with evidence quotes, every seller offer with buy_url), and shelf_context gives its rank on its category shelf by value and quality, its variants (colorways, sizes), and up to three higher-scoring alternatives in the same price band. A product with no live price or image is still returned, flagged under readiness; its price is not current. Responses that carry buy_url links also carry a disclosure field: the commission disclosure for those links.
brandquerycategoryidentifiersuser_signatureGet product details
Full detail for up to 25 products: all specifications in the category's standard order (card_specs first), review-derived ratings with verbatim evidence quotes, every seller offer with price and buy_url, and attribute_provenance recording where each spec value came from (manufacturer page, retailer listing, spec table). ids accepts product ids and shelf row ids alike: a group row's id resolves to the product the row is named after (its top_pick_id), and resolved_groups reports each substitution. Responses that carry buy_url links also carry a disclosure field: the commission disclosure for those links.
ids必填user_signature已滿足適用檢查項的 90%
該評分衡量此收錄條目的完整程度，依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價，也不表示伺服器安全或目前可用：可連通性是帶日期的檢查結果，評分也不包含安裝量或使用者評價。
請在儲存庫或註冊中心條目中宣告授權條款。
https://us-central1-swervie-prod.cloudfunctions.net/mcpclaude mcp add --transport http swervie https://us-central1-swervie-prod.cloudfunctions.net/mcpMCP=https://us-central1-swervie-prod.cloudfunctions.net/mcpH=(-H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream')
# Handshake: server info, capabilities, agent instructionscurl -s -X POST $MCP "${H[@]}" -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"curl","version":"1.0"}}}'
# A whole shopping answer in one callcurl -s -X POST $MCP "${H[@]}" -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"search_products","arguments":{"query":"65 inch OLED TV under $1500","limit":3}}}'user_signature = "sv1-" + sha256hex("swervie/v1|" + id).slice(0, 24)