为 agent 准备的选卡 oracle：此刻该用哪张信用卡付款？ 答案来自 RightCard 经银行发布、并由两个来源交叉验证的奖励数据——与 RightCard iOS app 使用的数据完全相同。
无需银行登录。不创建账户。不存储任何数据。 你只需传入卡片 id 和商店名称，就能立刻得到答案。
构建方式本身即诚实：带商户分类明细（如 Costco 是仓储会员店，因此杂货返现不会入账）、需自行激活的轮换季度、并列时的卡会备注出来，积分按换成某计划的 value。
默认为现金价值；若需要更保守的旅行价值，则传 valuation: "points"。
托管模式（无状态 Streamable HTTP）：
https://mcp.rightcard.ai/mcp本地模式（stdio，除从公共目录读取外，不会向外发送任何数据）：
npx rightcard-mcpClaude Desktop / Claude Code / Cursor config：
{ "mcpServers": { "rightcard": { "command": "npx", "args": ["-y", "rightcard-mcp"] } } }Related MCP server: Blackwell Bank Card Hub MCP
工具 | 说明 |
| 给定卡片中，今天在某个商店或类别赚得最多的是哪张，并带提示 |
| 商店被如何分类（类别、商户类型、MCC 避坑说明） |
| 按名称/发卡行查找 card id；已验证的排在前面，未验证的会被标记 |
| 某一张卡的已验证返现比例、轮换时段、可自选分类方式 |
| 当前与未来已开启的轮换季度，以及长期有效的商户权益 |
试试：“我持有 Amex Gold、Freedom KDR和 Citi Double Cash。在 Costco 应该用哪一张？”
src/engine.ts 是从 iOS App 的 Swift 引擎移植而来。fixtures/engine_golden.json 包含了 Swift 代码在真实卡片目录和商户名系列目录上产出的 3,100+ 答案；只要移植与它们其中任何一条不一致，npm test 都会失败。个性化的银行 Offer（用户个人专属银行优惠）刻意不包含在本服务中—它们仍保留在移动端。
托管服务只统计请求次数。不收集标识符，不使用 Cookie，不保存任何数据。完整隐私声明：https://rightcard.ai/privacy
Which of the given cards earns the most at a store (merchant) or for a spend category, today. Returns the card, the honest rate, why, and caveats (merchant-code traps, activation, ties). Base rates + rotating bonuses + built-in card benefits only; a user's personal bank offers live on their phone and are not consulted here.
Inputs: wallet (required), on, category, merchant, +2 more.
Everything RightCard knows about one card: rates by category, base rate, currency, annual fee, verification status, live rotating windows, and the choose-your-category spec if it has one.
Inputs: id (required).
The store's reward category as RightCard's directory files it, its merchant type, and the merchant-code caveat when a category bonus usually will not post there (superstores like Walmart/Target, warehouse clubs like Costco/Sam's). Same ranking the app uses.
Inputs: name (required).
Live and upcoming rotating-category windows (Freedom Flex, Discover it, Citi Dividend…) and permanent merchant benefits, from the same verified data as the app. Optionally filter by card id.
Inputs: card_id.
Search RightCard's catalog by card or issuer name to get the ids to pass into best_card. Verified cards first; unverified cards are flagged, never hidden.
Inputs: query (required), limit, issuer.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Scored across 5 tools
Each tool has a clear, distinct purpose: best_card compares cards, lookup_merchant provides merchant category info, search_cards finds card IDs, card gives detailed card data, and rotating_calendar lists rotating offers. There is no overlap or ambiguity between them.
All tool names use snake_case and are readable. Most follow verb_noun pattern (lookup_merchant, search_cards) but 'card' and 'rotating_calendar' differ slightly; still, the style is consistent and predictable.
With 5 tools, the server is well-scoped for its purpose of card rewards recommendations. Each tool is necessary and not redundant, fitting comfortably in the ideal 3-15 range.
The domain of finding the best card for a merchant or category is fully covered: search for cards, retrieve card details, check merchant categories, compare cards, and consult rotating offers. There are no obvious missing operations for this specific use case.
Search, compare and estimate rewards across US credit cards. Verified fees, APRs and bonuses.
AI credit card advisor - search cards, compare portfolios, and optimize rewards
Miles' public credit-card catalog: facts, transfer partners, valuations. Keyless, read-only.
Search live credit card merchant offers across Amex, Chase, Citi, PayPal, Rakuten, Capital One.
