carkeeper-mcp
Enables ChatGPT to run live CarKeeper vehicle checks, full purchase analyses, and multi-car comparisons using real report data, with optional paid report checkout tools.
README
CarKeeper ChatGPT MCP integration
This project connects ChatGPT to the live CarKeeper report engine.
What it exposes
check_vehicle— runs the real CarKeeper engine for a registration.analyse_purchase— runs the full CarKeeper buying report with price, advert text and optional images.compare_vehicles— runs CarKeeper on 2–5 cars and ranks them by the Buyer Score returned by the live report.start_full_report_checkout— optional/private tool that creates the existing £9.99 CarKeeper WooCommerce checkout.get_report— optional/private tool that reads status/text for an existing paid report.
The commerce tools are disabled by default because the current public Plugin Directory form says digital-goods commerce is not supported. They can still be enabled in Developer Mode for private testing.
Architecture
ChatGPT -> MCP server -> protected CarKeeper Render endpoint -> DVLA/DVSA/Check Car Details/web/OpenAI
Optional private purchase path: ChatGPT -> MCP server -> WordPress REST bridge -> existing WooCommerce checkout/report flow
1. Patch the Render report engine
Use the supplied carkeeper-report-engine-plugin version of the backend. It keeps /generate-report unchanged for the website and adds:
POST /plugin/generate-report
This route requires:
Authorization: Bearer <CARKEEPER_PLUGIN_TOKEN>
Add CARKEEPER_PLUGIN_TOKEN to the Render service environment.
2. Configure this MCP server
Copy .env.example to .env and set the SAME CARKEEPER_PLUGIN_TOKEN value.
For public/plugin testing, leave:
CARKEEPER_ENABLE_COMMERCE_TOOLS=false
3. Install and run
npm install
npm start
The MCP endpoint is:
http://localhost:8787/mcp
4. Test locally
npx @modelcontextprotocol/inspector@latest
Connect Inspector to http://localhost:8787/mcp using Streamable HTTP.
5. Connect to ChatGPT Developer Mode
Expose port 8787 using an HTTPS tunnel such as ngrok and use:
https://YOUR-TUNNEL/mcp
as the app's MCP URL.
Optional: existing £9.99 checkout flow
Install carkeeper-chatgpt-api.php as a small WordPress plugin. Add this constant somewhere secure in WordPress configuration:
define('CK_CHATGPT_API_TOKEN', 'a-long-random-secret');
Set the same value as CARKEEPER_WP_API_TOKEN on the MCP server and set:
CARKEEPER_ENABLE_COMMERCE_TOOLS=true
for private Developer Mode testing only.
The bridge uses the existing CarKeeper report creation, checkout token restoration, £9.99 WooCommerce product and paid-report storage rather than creating a second payment system.
Security
Never commit .env, OpenAI keys, Check Car Details keys, DVLA keys or shared plugin tokens. Generate long random secrets for both shared tokens.
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。