Periskop MCP Server

Periskop MCP Server

Enables AI agents to perform product discovery from natural language shopping intents, returning ranked products with merchant links without completing checkout.

Category
访问服务器

README

Periskop — Product Discovery for AI Agents

Product discovery for AI agents. One MCP call turns a natural-language shopping intent into structured, ranked product results — best picks, alternatives, bundles, caveats, prices, and merchant links — from across independent stores on the open merchant web.

Periskop is a hosted, remote MCP server (closed source). You connect over the public HTTPS endpoint with a Periskop API key — nothing to install or run. It's the discovery layer for agentic commerce: an agent asks in natural language and Periskop returns ranked products with links back to the merchant. It stops at discovery — no checkout, no payments, no orders.

  • Website: https://periskop.ai
  • Developer portal / get an API key: https://periskop.ai/developer
  • Docs: https://periskop.ai/developer/docs
  • MCP endpoint: https://mcp.periskop.ai/v1/mcp (Streamable HTTP, JSON-RPC 2.0)

Who it's for

Any product that needs to turn intent into reliable product results without owning checkout — agent platforms and orchestrators, AI worker platforms, AI shopping assistants, browser agents, commerce copilots, recommendation systems, procurement and B2B sourcing workflows, price and deal monitors, restock/reorder bots, gifting apps, resale and arbitrage workflows, and accessibility shopping tools. Concretely: an agent that plans a camping trip and returns the gear, a procurement copilot comparing options across suppliers, or a voice assistant surfacing the right product from a single sentence.

What it does (and does not) do

Periskop returns product results and merchant links only. It does not complete checkout, create merchant carts, process payment, reserve inventory, or purchase items. The user always completes any purchase on the merchant's own website. Every response carries a purchase_boundary block restating this.

Connection

Endpoint https://mcp.periskop.ai/v1/mcp
Transport Streamable HTTP — JSON-RPC 2.0 over HTTP POST (one JSON response per POST; no SSE)
Wire protocol 2024-11-05
Auth Authorization: Bearer <YOUR_PERISKOP_API_KEY> (keys look like dp_…)
Get a key https://periskop.ai/developer

Unauthenticated requests receive 401 with a WWW-Authenticate: Bearer header.

Note: Periskop also supports OAuth 2.1 (Authorization Code + PKCE) for hosts that prefer it (tokens look like dpo_…). API key is the simplest path and is all you need for the configs below.

Tools

All tools take a single JSON object argument. Schemas below are the public tool surface.

run_shopping_discovery

Find, choose, browse, recommend, get the best product, or build a bundle from a natural-language request. The only required field is prompt.

Field Type Required Notes
prompt string ✅ Natural-language shopping intent
mode string | null auto | browse | recommend | best | bundle (default auto)
store string | null auto, a store id, or a store name/hint
country string | null e.g. PT, ES
currency string | null e.g. EUR
language string | null e.g. en, pt
max_results integer | null Cap on returned products
response_format string | null full (default) | simple

get_discovery_result

Retrieve a previous result by result_id. Results are temporary and may expire.

Field Type Required
result_id string ✅
response_format string | null

discover_supported_stores

Inspect the public stores Periskop can use. Returns public store capabilities only.

Field Type Required
country string | null
category string | null
capability string | null (search | bundle | product_links)

report_result_feedback

Report whether a result was good, bad, or mixed.

Field Type Required
result_id string ✅
rating string (good | bad | mixed) ✅
reason string | null
selected_product_id string | null
freeform_feedback string | null

suggest_store_coverage

Suggest a store/merchant/marketplace Periskop should support in the future. Non-billable. Does not search or scrape the store and does not guarantee future support.

Field Type Required
store_name string ✅
store_url string | null
country / region / category string | null
context string | null (e.g. unsupported_store, no_match)

Example

Request (JSON-RPC tools/call)

curl -sS -X POST https://mcp.periskop.ai/v1/mcp \
  -H "Authorization: Bearer YOUR_PERISKOP_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "run_shopping_discovery",
      "arguments": { "prompt": "wireless noise-cancelling headphones for travel under 200€", "mode": "best" }
    }
  }'

Response (trimmed — sample/fake data, not real)

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "content": [
      {
        "type": "text",
        "text": "{\"result_id\":\"res_example_8f2a1c\",\"mode_used\":\"best\",\"products\":[{\"title\":\"SampleAudio Aero NC\",\"price\":\"€179.00\",\"currency\":\"EUR\",\"merchant\":\"Example Store\",\"url\":\"https://store.example/p/aero-nc\",\"role\":\"best_pick\"},{\"title\":\"Acme Quietline 2\",\"price\":\"€149.00\",\"currency\":\"EUR\",\"merchant\":\"Example Store\",\"url\":\"https://store.example/p/quietline-2\",\"role\":\"alternative\"}],\"caveats\":[\"Prices and availability may change on the merchant site.\"],\"purchase_boundary\":{\"checkout_created\":false,\"payment_processed\":false,\"stock_reserved\":false,\"user_must_complete_purchase_on_merchant_site\":true}}"
      }
    ]
  }
}

The structured payload is returned as text in result.content[0].text. Every response and the X-Request-ID response header carry a per-call request id.

Pricing (summary)

Prepaid wallet, billed only on successful discovery requests. Auth errors, malformed requests, internal/runtime failures, rate limits, and hard no-match are never billed. See the live rate at https://periskop.ai/developer/billing.

Client setup

Copy-paste configs are in config-snippets/: Cursor (.cursor/mcp.json), Claude Desktop, Claude Code, and a raw curl example.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选