portal-api
A read-only MCP server for Portal.io that enables browsing proposals, users, contacts, and catalog items through secure HMAC-signed API requests.
README
Portal.io API MCP
A local, read-only MCP server for a Portal.io account. It signs API requests with Portal's HMAC-SHA256 scheme and makes account data available to Codex through structured tools.
Included tools
test_connection: confirm the current credentials can make a read-only request.list_proposalsandget_proposal: browse proposal summaries and details.list_users: look up Portal salesperson IDs.search_contacts: find contacts in the Portal account.search_catalog: search account catalog items when Catalog API access is enabled.
Version 0.1.0 contains no mutation tools. It cannot create, update, archive, or send anything in Portal.
Install
git clone https://github.com/AWfusion/portal-api-mcp.git
cd portal-api-mcp
npm ci
cp .env.example .env
Fill in .env with credentials from the same Portal environment:
PORTAL_ENV=sandbox
PORTAL_BASE_URL=https://sandbox.api.portal.io
PORTAL_APP_ID=your-api-application-key
PORTAL_SECRET_KEY=your-secret-key
PORTAL_USER_API_KEY=the-meta-apiKey-value-from-the-credential-exchange
PORTAL_WRITE_ENABLED=false
For production, use PORTAL_ENV=production and PORTAL_BASE_URL=https://api.portal.io only when Portal has issued and enabled production credentials for the integration.
PORTAL_USER_API_KEY must be the exact meta.apiKey returned by GET /authenticate/apikeyexchange. Do not substitute the application ID, secret key, a Portal UI token, or the username/password used for the one-time exchange. Portal's returned User API Key is Base64-encoded and is normally 44 characters long.
The actual .env file is ignored by Git. Keep its permissions owner-only:
chmod 600 .env
Verify
npm test
npm run verify
npm run verify performs a local MCP initialization, confirms every tool is registered, and makes one read-only GET /public/proposals?PageNumber=1&PageSize=1 call. It outputs no credentials.
Register Globally in Codex
Run the included installer from the project directory:
./scripts/install-codex-mcp.sh
It registers this local process globally under the name portal-api. Confirm it with:
codex mcp get portal-api
Restart Codex or open a new task after registration so the tools are discovered. Typical requests include “list my most recently modified Portal proposals” and “search the catalog for Denon receivers.”
Troubleshooting
app id could not be found: the API application ID is not enabled for the selected base URL. Sandbox credentials belong athttps://sandbox.api.portal.io; production credentials belong athttps://api.portal.io.request signature (hash) is invalid: first confirmPORTAL_USER_API_KEYis the Base64meta.apiKeyreturned by the credential exchange, then confirm the app ID, secret, and User API Key all came from the same environment.- Catalog queries return authorization errors: ask Portal to enable Catalog API access for the account.
Development
The server uses Node 20 or newer and has no database or hosted component. Request signing follows Portal's current HMAC documentation: GET signatures omit query parameters and content type; the API secret is used as the raw ASCII HMAC key and is not Base64-decoded.
See CHANGELOG.md for release history and BUILD-INFO.md for the release contract.
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。